flow: reference host driver flow-drive-host/flow-run-host + 4 tests

Completes the host ABI from work-queue to driver loop: the host supplies only a
(kind payload) -> answer dispatch fn; flow-drive-host services one tick of pending
requests, flow-run-host ticks until quiescent (bounded). Tested via the art-dag
render -> human-review -> publish pipeline driven entirely by flow-run-host. The
art-dag integration is now: define dispatch, call flow-run-host. 166/166, 11 suites.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

lib/flow/README.md

@@ -0,0 +1,141 @@
+# flow — durable DAG workflows on Scheme
+
+`flow` is a workflow engine for rose-ash: content pipelines (write → review →
+publish → federate), scheduled jobs, and multi-step user flows (signup, confirm,
+onboard) that **survive process restarts**. It is a thin Scheme prelude over the
+Scheme-on-SX guest (`lib/scheme/`); a flow runs *inside* the interpreter.
+
+Run the suite: `bash lib/flow/conformance.sh` → **151/151 across 10 suites**.
+
+## Model
+
+A **flow** is just a Scheme procedure of one argument — the upstream value:
+
+```
+node : input -> output
+```
+
+Combinators build composite nodes out of child nodes. A node that ignores its
+argument is effectively a thunk. There is no separate "graph" object: composition
+*is* function composition, so flows are values you can name, pass, and nest.
+
+```scheme
+(defflow publish
+  (sequence
+    (lambda (draft) (string-append draft "!"))
+    (branch (lambda (post) (>= (string-length post) 3))
+            (remote-node 'fed 'publish)
+            (flow-const 'rejected))))
+
+(flow/start publish "hello")   ; => federated, or a (flow-suspended id tag) state
+```
+
+## Building blocks (`spec.sx`)
+
+| Combinator | Meaning |
+|---|---|
+| `(flow-node f)` / `(flow-id x)` / `(flow-const v)` | leaf nodes |
+| `(sequence n ...)` | thread input left-to-right |
+| `(parallel n ...)` | fan input to every child, join results into a list (sequential eval) |
+| `(map-flow node)` | run `node` over each item of a list input, join results |
+| `(flow-while pred body max)` / `(flow-until ...)` | bounded iteration (cap `max` steps) |
+| `(defflow name body)` | bind + register a named flow (so it survives restart) |
+
+## Control flow + errors (`spec.sx`)
+
+| Combinator | Meaning |
+|---|---|
+| `(branch pred then else)` | `pred` on input selects `then`/`else` (`cond` is a Scheme special form) |
+| `(retry n node)` | re-run on a *raised exception*, up to `n` attempts |
+| `(timeout budget node)` | cooperative **step budget**: nodes call `(tick)`; the `(budget+1)`-th tick raises `flow-timeout` |
+| `(try-catch node handler)` | catch a raised exception → `(handler error)` |
+| `(fail reason)` / `(failed? x)` / `(fail-reason x)` | explicit failure *values* (flow downstream as data) |
+| `(recover node handler)` | the fail-VALUE counterpart of try-catch |
+| `(attempt n ...)` | railway sequence: stop at the first node returning a `(fail ...)` |
+| `(tap effect)` | run a side effect, return input unchanged |
+
+**Two error channels, on purpose.** Raised exceptions are for *bugs/transients*
+(caught by `retry`/`try-catch`). `(fail reason)` values are for *expected business
+outcomes* (validation rejected, declined) and compose via `attempt`/`recover`.
+
+## Suspend / resume — the durable core (`spec.sx`, `store.sx`)
+
+The guest Scheme's `call/cc` is **escape-only** — re-invoking a captured
+continuation after it returns *hangs* the runtime. So flow does **not** serialize
+continuations. Instead it uses **deterministic replay**:
+
+- `(suspend tag)` — if `tag` is already in the replay log, return its logged value;
+  otherwise escape to the driver as `(flow-suspended tag)`.
+- `resume` appends `(tag value)` to the log and **re-runs the flow from the start**.
+  Already-resolved suspends replay their values; the first unresolved one escapes
+  again (or the flow completes).
+
+The entire persisted state is the replay log — plain data. No live continuation is
+ever stored, so flows survive process restarts and even moves between instances.
+
+> **Author contract:** suspend `tag`s must be unique and deterministic across
+> replays, and **all** non-determinism / side effects must go through suspend
+> points (so their results are logged) — otherwise they re-run on every replay.
+
+### Lifecycle (`store.sx`)
+
+```scheme
+(flow/start flow input)   ; raw result if it completes, else (flow-suspended id tag)
+(flow/resume id value)    ; inject value at the waiting tag, continue
+(flow/cancel id)          ; terminate; a later resume is rejected
+```
+
+### Introspection & hygiene
+
+```scheme
+(flow/status id)   ; done | suspended | cancelled | unknown
+(flow/result id)   ; result if done, else (flow-error reason)
+(flow/list)        ; ((id status) ...)
+(flow/pending)     ; ((id waiting-tag) ...) — what each suspended flow awaits
+(flow/gc)          ; drop terminal records, keep live ones; returns count removed
+(flow/forget id)   ; drop one terminal record (refuses live flows)
+```
+
+### Crash recovery
+
+```scheme
+(flow-store-export)      ; the store as plain data (live procs nulled)
+(flow-store-import! d)   ; restore the store from exported data
+(flow-resumable-ids)     ; ids of suspended flows to wake on restart
+```
+
+On restart the flow definitions are reloaded (`defflow` re-registers names) and the
+exported store reimported; `resume` re-resolves each flow's procedure **by name**.
+
+## Distribution via fed-sx (`remote.sx`)
+
+```scheme
+(flow-peer-register! addr table)      ; mock a peer's exposed functions (fed-sx boundary)
+(remote-node addr fn)                 ; run a node on a peer
+(remote-failover addrs fn local)      ; try peers in order, fall through to a local node
+(flow-replicate-to addr)              ; copy this store to a peer's replica slot
+(flow-restore-from addr)              ; import a peer's replica (handoff)
+```
+
+**Handoff** is crash recovery across instances: replicate → local instance dies →
+peer restores the (plain-data) store and resumes. The replay log carries over, so
+all resolved suspends survive the move.
+
+## Files
+
+| File | Contents |
+|---|---|
+| `spec.sx` | combinators (flow-combinators-src / flow-control-src / flow-suspend-src) |
+| `store.sx` | durable store, lifecycle, crash recovery, introspection, hygiene |
+| `remote.sx` | fed-sx transport (mock peer registry), failover, replication |
+| `api.sx` | `flow-make-env` / `flow-run` SX helpers (one cached env, per-test reset) |
+| `tests/*.sx` | 10 suites, 151 cases |
+| `conformance.sh` | loads substrate + flow layer, runs every suite |
+
+## Notes on the substrate
+
+The guest Scheme (`lib/scheme/`, imported read-only) lacks dotted-rest params
+`(a . rest)` and named `let`; combinators use `(lambda args ...)` variadics + top-
+level recursion. `cons` is list-only (no dotted pairs), so log/assoc entries are
+2-element lists. Strings box as `{:scm-string "..."}`. Timeout is a step budget
+because there is no wall clock; `parallel` is sequential for the same reason.

lib/flow/api.sx

@@ -0,0 +1,65 @@
+;; lib/flow/api.sx — flow runtime entry points.
+;;
+;; Builds a Scheme env preloaded with the flow combinators (lib/flow/spec.sx),
+;; the durable store + lifecycle (lib/flow/store.sx), the fed-sx remote layer
+;; (lib/flow/remote.sx), and the host integration ABI (lib/flow/host.sx), and
+;; provides SX helpers to run flow programs.
+;;
+;; Scheme-level API (available inside flow programs):
+;;   (flow/start flow input) — run a flow; raw result if it completes, else
+;;                             (flow-suspended id tag). Defined in store.sx.
+;;   (flow/resume id value)  — resume a suspended flow (store.sx)
+;;   (flow/cancel id)        — cancel a flow (store.sx)
+;;   (suspend tag)           — suspension point (spec.sx)
+;;   (request kind payload)  — host request envelope over suspend (host.sx)
+;;   (remote-node addr fn)   — node executed on a federation peer (remote.sx)
+;;
+;; SX-level helpers (for hosts and tests):
+;;   (flow-make-env)        — fresh standard env + combinators + store + remote + host
+;;   (flow-run src)         — eval a Scheme program string in a reset shared env
+;;   (flow-run-in env src)  — eval a Scheme program string in a given env
+;;
+;; flow-run reuses ONE env (building the full standard env is expensive) and
+;; resets the mutable flow globals before each program, so tests stay isolated
+;; without paying for a fresh standard env each time. flow-registry persists (it
+;; models reloaded flow definitions surviving a restart).
+
+(define
+  flow-make-env
+  (fn
+    ()
+    (let
+      ((env (scheme-standard-env)))
+      (flow-load-combinators! env)
+      (flow-load-store! env)
+      (flow-load-remote! env)
+      (flow-load-host! env)
+      env)))
+
+(define
+  flow-run-in
+  (fn (env src) (scheme-eval-program (scheme-parse-all src) env)))
+
+(define
+  flow-reset-src
+  "(set! flow-store (list)) (set! flow-next-id 0) (set! flow-replay-log (list)) (set! flow-suspend-k #f) (set! flow-timeout-budget -1) (set! flow-peers (list)) (set! flow-replicas (list))")
+
+(define flow-env-cache false)
+
+(define
+  flow-shared-env
+  (fn
+    ()
+    (begin
+      (if flow-env-cache nil (set! flow-env-cache (flow-make-env)))
+      flow-env-cache)))
+
+(define
+  flow-run
+  (fn
+    (src)
+    (let
+      ((env (flow-shared-env)))
+      (begin
+        (scheme-eval-program (scheme-parse-all flow-reset-src) env)
+        (scheme-eval-program (scheme-parse-all src) env)))))

lib/flow/conformance.sh

@@ -0,0 +1,103 @@
+#!/usr/bin/env bash
+# flow-on-sx conformance runner — runs all flow test suites in one sx_server process.
+#
+# Usage:
+#   bash lib/flow/conformance.sh        # run all suites
+#   bash lib/flow/conformance.sh -v     # verbose (list each suite)
+
+set -uo pipefail
+cd "$(git rev-parse --show-toplevel)"
+
+SX_SERVER="${SX_SERVER:-hosts/ocaml/_build/default/bin/sx_server.exe}"
+if [ ! -x "$SX_SERVER" ]; then
+  SX_SERVER="/root/rose-ash/hosts/ocaml/_build/default/bin/sx_server.exe"
+fi
+if [ ! -x "$SX_SERVER" ]; then
+  echo "ERROR: sx_server.exe not found." >&2
+  exit 1
+fi
+
+VERBOSE="${1:-}"
+
+# Suites: NAME RUNNER-FN PATH
+SUITES=(
+  "basic       flow-basic-tests-run!     lib/flow/tests/basic.sx"
+  "control     flow-ctl-tests-run!       lib/flow/tests/control.sx"
+  "suspend     flow-sus-tests-run!       lib/flow/tests/suspend.sx"
+  "recovery    flow-rec-tests-run!       lib/flow/tests/recovery.sx"
+  "distributed flow-dist-tests-run!      lib/flow/tests/distributed.sx"
+  "api         flow-api-tests-run!       lib/flow/tests/api.sx"
+  "combinators flow-cmb-tests-run!       lib/flow/tests/combinators.sx"
+  "railway     flow-rail-tests-run!      lib/flow/tests/railway.sx"
+  "integration flow-int-tests-run!       lib/flow/tests/integration.sx"
+  "hygiene     flow-hyg-tests-run!       lib/flow/tests/hygiene.sx"
+  "host        flow-hst-tests-run!       lib/flow/tests/host.sx"
+)
+
+TMPFILE=$(mktemp); trap "rm -f $TMPFILE" EXIT
+EPOCH=1
+
+emit_load () { echo "(epoch $EPOCH)"; echo "(load \"$1\")"; EPOCH=$((EPOCH+1)); }
+emit_eval () { echo "(epoch $EPOCH)"; echo "(eval \"$1\")"; EPOCH=$((EPOCH+1)); }
+
+{
+  emit_load "lib/guest/lex.sx"
+  emit_load "lib/guest/reflective/env.sx"
+  emit_load "lib/guest/reflective/quoting.sx"
+  emit_load "lib/scheme/parser.sx"
+  emit_load "lib/scheme/eval.sx"
+  emit_load "lib/scheme/runtime.sx"
+  emit_load "lib/flow/spec.sx"
+  emit_load "lib/flow/store.sx"
+  emit_load "lib/flow/remote.sx"
+  emit_load "lib/flow/host.sx"
+  emit_load "lib/flow/api.sx"
+  for SUITE in "${SUITES[@]}"; do
+    read -r _NAME _RUNNER FILE <<< "$SUITE"
+    emit_load "$FILE"
+    emit_eval "($_RUNNER)"
+  done
+} > "$TMPFILE"
+
+OUTPUT=$(timeout 540 "$SX_SERVER" < "$TMPFILE" 2>&1 || true)
+
+TOTAL_PASS=0
+TOTAL_FAIL=0
+FAILED_SUITES=()
+
+LAST_DICT_LINES=$(echo "$OUTPUT" | grep -E '^\{:' || true)
+
+I=0
+while read -r LINE; do
+  [ -z "$LINE" ] && continue
+  P=$(echo "$LINE" | grep -oE ':passed [0-9]+' | awk '{print $2}')
+  F=$(echo "$LINE" | grep -oE ':failed [0-9]+' | awk '{print $2}')
+  [ -z "$P" ] && P=0
+  [ -z "$F" ] && F=0
+  SUITE_INFO="${SUITES[$I]}"
+  SUITE_NAME=$(echo "$SUITE_INFO" | awk '{print $1}')
+  TOTAL_PASS=$((TOTAL_PASS + P))
+  TOTAL_FAIL=$((TOTAL_FAIL + F))
+  if [ "$F" -gt 0 ]; then
+    FAILED_SUITES+=("$SUITE_NAME: $P/$((P+F))")
+    printf 'X %-12s %d/%d\n' "$SUITE_NAME" "$P" "$((P+F))"
+    echo "$LINE" | grep -oE ':name "[^"]*"' | sed 's/:name /    fail: /'
+  elif [ "$VERBOSE" = "-v" ]; then
+    printf 'ok %-12s %d passed\n' "$SUITE_NAME" "$P"
+  fi
+  I=$((I+1))
+done <<< "$LAST_DICT_LINES"
+
+TOTAL=$((TOTAL_PASS + TOTAL_FAIL))
+if [ "$TOTAL" -eq 0 ]; then
+  echo "ERROR: no suite results parsed. Raw output:" >&2
+  echo "$OUTPUT" >&2
+  exit 1
+fi
+if [ $TOTAL_FAIL -eq 0 ]; then
+  echo "ok $TOTAL_PASS/$TOTAL flow-on-sx tests passed (${#SUITES[@]} suites)"
+else
+  echo "FAIL $TOTAL_PASS/$TOTAL passed, $TOTAL_FAIL failed:"
+  for S in "${FAILED_SUITES[@]}"; do echo "  $S"; done
+  exit 1
+fi

lib/flow/host.sx

@@ -0,0 +1,42 @@
+;; lib/flow/host.sx — the host integration ABI (Phase 8).
+;;
+;; `suspend` is flow's seam to the outside world, but a bare (suspend tag) is just a
+;; signal — every author would invent their own tag shape. This layer defines a
+;; stable request/response contract so a host (e.g. an art-dag driver, or a human
+;; review UI) can hook in WITHOUT reverse-engineering ad-hoc tags.
+;;
+;; A flow asks the host to do something and waits for the answer:
+;;   (request kind payload)  — suspend with a typed envelope (flow-request kind
+;;     payload); evaluates to the host's resume value.
+;;   (await-human prompt)    — request kind=human (a decision point)
+;;   (await-render recipe)   — request kind=render (e.g. an art-dag job)
+;;   (await-effect kind p)   — request of an arbitrary kind
+;;
+;; The host drives flows by polling its work queue and resuming:
+;;   (flow-host-requests)    — ((id kind payload) ...) for every SUSPENDED flow whose
+;;     waiting tag is a host request. The host dispatches by kind (render -> submit a
+;;     Celery job; human -> show UI), then calls (flow/resume id answer).
+;;   (request? tag) / (request-kind tag) / (request-payload tag) — parse one tag.
+;;
+;; Reference driver — the host only supplies `dispatch`, a (kind payload) -> answer:
+;;   (flow-drive-host dispatch)        — one tick: service every CURRENTLY pending
+;;     request (snapshot), resuming each with (dispatch kind payload); returns the
+;;     count serviced. Resumes may create new requests — serviced on the next tick.
+;;   (flow-run-host dispatch maxticks) — tick until quiescent (no pending requests)
+;;     or maxticks reached; returns total requests serviced. Bounded for determinism.
+;;
+;; Contract: the host owns IO and persistence. flow stays deterministic — a flow
+;; never performs IO itself, it only `request`s; the host performs the effect and
+;; feeds the result back via resume (which the replay log records, so the effect is
+;; not re-run on recovery). Persist with flow-store-export after each transition and
+;; flow-store-import! on boot.
+
+(define
+  flow-host-src
+  "(define (request kind payload) (suspend (list (quote flow-request) kind payload)))\n   (define (request? tag) (and (pair? tag) (eq? (car tag) (quote flow-request))))\n   (define (request-kind tag) (car (cdr tag)))\n   (define (request-payload tag) (car (cdr (cdr tag))))\n   (define (await-human prompt) (request (quote human) prompt))\n   (define (await-render recipe) (request (quote render) recipe))\n   (define (await-effect kind payload) (request kind payload))\n   (define (flow-host-req-step pend)\n     (if (null? pend)\n         (list)\n         (let ((id (car (car pend))) (tag (car (cdr (car pend)))))\n           (if (request? tag)\n               (cons (list id (request-kind tag) (request-payload tag))\n                     (flow-host-req-step (cdr pend)))\n               (flow-host-req-step (cdr pend))))))\n   (define (flow-host-requests) (flow-host-req-step (flow/pending)))\n   (define (flow-drive-host-step reqs dispatch)\n     (if (null? reqs)\n         0\n         (begin\n           (flow/resume (car (car reqs)) (dispatch (car (cdr (car reqs))) (car (cdr (cdr (car reqs))))))\n           (+ 1 (flow-drive-host-step (cdr reqs) dispatch)))))\n   (define (flow-drive-host dispatch) (flow-drive-host-step (flow-host-requests) dispatch))\n   (define (flow-run-host dispatch maxticks)\n     (if (<= maxticks 0)\n         0\n         (let ((n (flow-drive-host dispatch)))\n           (if (= n 0) 0 (+ n (flow-run-host dispatch (- maxticks 1)))))))")
+
+(define
+  flow-load-host!
+  (fn
+    (env)
+    (begin (scheme-eval-program (scheme-parse-all flow-host-src) env) env)))

lib/flow/remote.sx

@@ -0,0 +1,34 @@
+;; lib/flow/remote.sx — distributed nodes via fed-sx (Phase 4).
+;;
+;; A node can execute on a federation peer. The transport is the fed-sx boundary;
+;; it is MOCKED in tests by a peer registry mapping addr -> function table. In
+;; production flow-transport would issue a fed-sx call; here it dispatches locally.
+;;
+;;   (flow-peer-register! addr table) — register a mock peer. table is a list of
+;;     (fn-name proc) entries — the functions that peer exposes.
+;;   (flow-transport addr fn input)   — invoke fn on the peer with input. Raises
+;;     (flow-remote-unreachable) if the addr is unknown, (flow-remote-no-fn) if the
+;;     peer does not expose fn.
+;;   (remote-node addr fn)            — a node that runs fn on the peer at addr.
+;;   (remote-failover addrs fn local) — try fn on each peer in addrs in order; on a
+;;     raised error move to the next peer; if every peer fails, run the `local`
+;;     node as a fallback.
+;;
+;; Persistence across instances + handoff. Each instance runs the same flow
+;; definitions, so the only thing that needs to cross the wire is the (plain-data)
+;; store — exactly flow-store-export from store.sx. Replication pushes that export
+;; to a peer's replica slot; handoff = restore the replica on the peer and resume.
+;;
+;;   (flow-replicate-to addr) — copy this instance's store to peer addr's replica
+;;   (flow-restore-from addr) — import the replica from peer addr (#t / #f)
+;;   (flow-replica-get addr)  — the raw replicated store at addr (or #f)
+
+(define
+  flow-remote-src
+  "(define flow-peers (list))\n   (define (flow-assoc key alist)\n     (if (null? alist)\n         #f\n         (if (eq? (car (car alist)) key) (car (cdr (car alist))) (flow-assoc key (cdr alist)))))\n   (define (flow-peer-register! addr table) (set! flow-peers (cons (list addr table) flow-peers)))\n   (define (flow-transport addr fn input)\n     (let ((table (flow-assoc addr flow-peers)))\n       (if table\n           (let ((proc (flow-assoc fn table)))\n             (if proc (proc input) (raise (quote flow-remote-no-fn))))\n           (raise (quote flow-remote-unreachable)))))\n   (define (remote-node addr fn) (lambda (input) (flow-transport addr fn input)))\n   (define (flow-failover-step addrs fn input local)\n     (if (null? addrs)\n         (local input)\n         (guard (e (#t (flow-failover-step (cdr addrs) fn input local)))\n           (flow-transport (car addrs) fn input))))\n   (define (remote-failover addrs fn local)\n     (lambda (input) (flow-failover-step addrs fn input local)))\n\n   (define flow-replicas (list))\n   (define (flow-replicas-remove addr reps)\n     (if (null? reps)\n         (list)\n         (if (eq? (car (car reps)) addr)\n             (flow-replicas-remove addr (cdr reps))\n             (cons (car reps) (flow-replicas-remove addr (cdr reps))))))\n   (define (flow-replicate-to addr)\n     (set! flow-replicas (cons (list addr (flow-store-export)) (flow-replicas-remove addr flow-replicas))))\n   (define (flow-replica-get addr) (flow-assoc addr flow-replicas))\n   (define (flow-restore-from addr)\n     (let ((data (flow-replica-get addr)))\n       (if data (begin (flow-store-import! data) #t) #f)))")
+
+(define
+  flow-load-remote!
+  (fn
+    (env)
+    (begin (scheme-eval-program (scheme-parse-all flow-remote-src) env) env)))

lib/flow/scoreboard.json

@@ -0,0 +1,19 @@
+{
+  "total": 166,
+  "passed": 166,
+  "failed": 0,
+  "suites": {
+    "basic": { "passed": 18, "total": 18 },
+    "control": { "passed": 31, "total": 31 },
+    "suspend": { "passed": 17, "total": 17 },
+    "recovery": { "passed": 8, "total": 8 },
+    "distributed": { "passed": 19, "total": 19 },
+    "api": { "passed": 12, "total": 12 },
+    "combinators": { "passed": 17, "total": 17 },
+    "railway": { "passed": 10, "total": 10 },
+    "integration": { "passed": 10, "total": 10 },
+    "hygiene": { "passed": 9, "total": 9 },
+    "host": { "passed": 15, "total": 15 }
+  },
+  "phases": { "phase1": "done", "phase2": "done", "phase3": "done", "phase4": "done", "phase5": "done", "phase6": "done", "phase7": "done", "phase8": "done" }
+}

lib/flow/scoreboard.md

@@ -0,0 +1,53 @@
+# flow-on-sx Scoreboard
+
+**All tests pass: 166 / 166 across 11 suites. Phases 1-8 complete.**
+
+`bash lib/flow/conformance.sh`
+
+## Per-suite breakdown
+
+| Suite | Passing | Covers |
+|-------|--------:|--------|
+| basic |     18  | Phase 1: single nodes, linear sequence, data-flow threading, defflow, parallel fan/join, nested composition, publish-shaped flow |
+| control |  31  | Phase 2: `branch` (6); error model `fail`/`failed?`/`fail-reason` (6); `try-catch` (6); `retry n` (6); `timeout` cooperative step budget (7) |
+| suspend |  17  | Phase 3: suspend/resume/cancel via deterministic replay; multi-step, replay determinism, lifecycle guards, suspend-in-branch |
+| recovery |  8  | Phase 3: crash recovery — store export/import, resumable scan, restart-at-every-step, replay-log survival |
+| distributed | 19 | Phase 4: `remote-node` (7); `remote-failover` (6); replication + handoff across instances (6) |
+| api |  12  | Phase 5: introspection — `flow/status`, `flow/result`, `flow/list`, `flow/pending` |
+| combinators | 17 | Phase 5: `tap`, `recover` (fail-value), `map-flow` fan-over-list, `flow-while`/`flow-until` bounded iteration |
+| railway | 10 | Phase 6: `attempt` — fail-value short-circuiting sequence + recover rejoin |
+| integration | 10 | Phase 7: end-to-end order + onboarding flows composing every phase (suspend, branch, federation, crash recovery, handoff, introspection) |
+| hygiene | 9 | Phase 5: `flow/gc` (prune terminal flows), `flow/forget` (drop one terminal record) |
+| host | 15 | Phase 8: host ABI — `request`/`await-human`/`await-render`, `flow-host-requests` queue, `flow-run-host` reference driver; art-dag-shaped render→review→publish loop |
+
+## Architecture
+
+Flow combinators are a **Scheme prelude** (`lib/flow/spec.sx`) loaded onto
+`scheme-standard-env`. A flow is a Scheme procedure `input -> output`. The whole
+flow executes inside the Scheme interpreter, so Phase 3's `suspend` (call/cc) will
+capture the flow continuation directly.
+
+- `lib/flow/spec.sx` — combinators: `flow-node`, `flow-id`, `flow-const`,
+  `sequence`, `parallel`, `defflow`; `flow-load-combinators!`.
+- `lib/flow/api.sx` — `flow/start` (Scheme); `flow-make-env`, `flow-run`,
+  `flow-run-in` (SX helpers).
+- `lib/flow/tests/basic.sx` — 18 cases.
+- `lib/flow/conformance.sh` — loads substrate + flow layer, runs suites.
+
+## Semantics notes
+
+- **node** = 1-arg Scheme procedure; the upstream value is the argument. A node
+  ignoring its argument is effectively a thunk.
+- **sequence** threads left-to-right; empty sequence = identity.
+- **parallel** fans the same input to every branch and joins results into a list.
+  Evaluation is **sequential** for now; true concurrency arrives in Phase 3.
+
+## Phases
+
+- [x] Phase 1 — Declarative DAG + sequential execution (combinators + 18 tests, `flow/start`)
+- [x] Phase 2 — Control flow + error handling (branch, error model, try-catch, retry, timeout)
+- [x] Phase 3 — Suspend/resume (suspend/resume/cancel + crash recovery via deterministic replay)
+- [x] Phase 4 — Distributed nodes via fed-sx (remote-node, failover, replication + handoff)
+- [x] Phase 5 — Operational API + combinators (introspection, tap, recover, map-flow)
+- [ ] Phase 3 — Suspend / resume (the showcase)
+- [ ] Phase 4 — Distributed nodes via fed-sx

lib/flow/spec.sx

@@ -0,0 +1,61 @@
+;; lib/flow/spec.sx — flow combinators as a Scheme prelude.
+;;
+;; A flow is a Scheme procedure of one argument: the upstream value.
+;;   node : input -> output
+;; A leaf node ignoring its argument is effectively a thunk. Combinators
+;; build composite nodes out of child nodes. The whole flow runs INSIDE the
+;; Scheme interpreter.
+;;
+;; Phase 1 combinators (flow-combinators-src):
+;;   flow-node / flow-id / flow-const / sequence / parallel / defflow
+;;   defflow both binds the flow and registers it by name (flow-register!, in
+;;   store.sx) so it can be re-resolved after a process restart.
+;;   map-flow (Phase 5): run a node over each item of a list input, join results.
+;;   flow-while / flow-until (Phase 5): bounded iteration — re-run body, threading
+;;   the value, while/until pred holds, up to `max` steps (deterministic bound; no
+;;   unbounded loops in pure SX).
+;;
+;; Phase 2 combinators (flow-control-src):
+;;   branch / fail / failed? / fail-reason / try-catch / retry / timeout / tick
+;;   tap (Phase 5): side-effecting pass-through (returns input unchanged).
+;;   recover (Phase 5): the fail-VALUE counterpart of try-catch.
+;;   attempt (Phase 6): railway sequence — thread nodes left-to-right but stop at
+;;     the first node that returns a (fail ...) value, returning that failure.
+;;
+;; Phase 3 suspend core (flow-suspend-src):
+;;   The guest Scheme's call/cc is ESCAPE-ONLY (re-invoking a captured k after it
+;;   returns hangs the runtime), so suspend/resume CANNOT re-enter a continuation.
+;;   Instead, durability uses DETERMINISTIC REPLAY: a flow re-runs from the start
+;;   on each resume; suspend points that have already been resolved replay their
+;;   logged value, and the first unresolved suspend escapes back to the driver.
+;;   The entire persisted state is the replay log (plain (tag value) data), which
+;;   survives process restart — no live continuation is ever serialized.
+;;
+;;   (suspend tag)  — if tag is in the replay log, return its value; else escape
+;;     to the driver as (flow-suspended tag). tags must be unique & deterministic
+;;     across replays. ALL effects/non-determinism must go through suspend so their
+;;     results are logged (otherwise they re-run on every replay).
+;;   (flow-drive flow input log) — run flow with the given replay log; returns
+;;     (flow-done result) or (flow-suspended tag).
+
+(define
+  flow-combinators-src
+  "(define (flow-node f) f)\n   (define (flow-id input) input)\n   (define (flow-const v) (lambda (input) v))\n   (define (flow-seq-step ns v)\n     (if (null? ns) v (flow-seq-step (cdr ns) ((car ns) v))))\n   (define sequence (lambda ns (lambda (input) (flow-seq-step ns input))))\n   (define parallel (lambda ns (lambda (input) (map (lambda (n) (n input)) ns))))\n   (define (map-flow node) (lambda (items) (map node items)))\n   (define (flow-while-step pred body input n)\n     (if (<= n 0)\n         input\n         (if (pred input) (flow-while-step pred body (body input) (- n 1)) input)))\n   (define (flow-while pred body max) (lambda (input) (flow-while-step pred body input max)))\n   (define (flow-until-step pred body input n)\n     (if (<= n 0)\n         input\n         (if (pred input) input (flow-until-step pred body (body input) (- n 1)))))\n   (define (flow-until pred body max) (lambda (input) (flow-until-step pred body input max)))\n   (define-syntax defflow\n     (syntax-rules ()\n       ((defflow nm body)\n        (begin (define nm body) (flow-register! (quote nm) nm)))))")
+
+(define
+  flow-control-src
+  "(define (branch pred then else)\n     (lambda (input) (if (pred input) (then input) (else input))))\n   (define (fail reason) (list (quote flow-fail) reason))\n   (define (failed? x) (and (pair? x) (eq? (car x) (quote flow-fail))))\n   (define (fail-reason x) (car (cdr x)))\n   (define (recover node handler)\n     (lambda (input)\n       (let ((r (node input)))\n         (if (failed? r) (handler (fail-reason r)) r))))\n   (define (tap effect)\n     (lambda (input) (begin (effect input) input)))\n   (define (flow-attempt-step ns v)\n     (if (failed? v)\n         v\n         (if (null? ns) v (flow-attempt-step (cdr ns) ((car ns) v)))))\n   (define attempt (lambda ns (lambda (input) (flow-attempt-step ns input))))\n   (define (try-catch node handler)\n     (lambda (input) (guard (e (#t (handler e))) (node input))))\n   (define (flow-retry-step n node input)\n     (guard (e (#t (if (<= n 1) (raise e) (flow-retry-step (- n 1) node input))))\n       (node input)))\n   (define (retry n node) (lambda (input) (flow-retry-step n node input)))\n   (define flow-timeout-budget -1)\n   (define (tick)\n     (if (< flow-timeout-budget 0)\n         0\n         (begin\n           (set! flow-timeout-budget (- flow-timeout-budget 1))\n           (if (< flow-timeout-budget 0)\n               (raise (quote flow-timeout))\n               flow-timeout-budget))))\n   (define (timeout budget node)\n     (lambda (input)\n       (let ((saved flow-timeout-budget))\n         (set! flow-timeout-budget budget)\n         (guard (e (#t (begin (set! flow-timeout-budget saved) (raise e))))\n           (let ((result (node input)))\n             (set! flow-timeout-budget saved)\n             result)))))")
+
+(define
+  flow-suspend-src
+  "(define flow-replay-log (list))\n   (define flow-suspend-k #f)\n   (define (flow-log-lookup tag log)\n     (if (null? log)\n         (list #f #f)\n         (if (eq? (car (car log)) tag)\n             (list #t (car (cdr (car log))))\n             (flow-log-lookup tag (cdr log)))))\n   (define (suspend tag)\n     (let ((hit (flow-log-lookup tag flow-replay-log)))\n       (if (car hit)\n           (car (cdr hit))\n           (flow-suspend-k (list (quote flow-suspended) tag)))))\n   (define (flow-drive flow input log)\n     (set! flow-replay-log log)\n     (call/cc\n       (lambda (k)\n         (set! flow-suspend-k k)\n         (list (quote flow-done) (flow input)))))")
+
+(define
+  flow-load-combinators!
+  (fn
+    (env)
+    (begin
+      (scheme-eval-program (scheme-parse-all flow-combinators-src) env)
+      (scheme-eval-program (scheme-parse-all flow-control-src) env)
+      (scheme-eval-program (scheme-parse-all flow-suspend-src) env)
+      env)))

lib/flow/tests/api.sx

@@ -0,0 +1,79 @@
+;; lib/flow/tests/api.sx — Phase 5: operational introspection API.
+
+(define flow-api-pass 0)
+(define flow-api-fail 0)
+(define flow-api-fails (list))
+
+(define
+  flow-api-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-api-pass (+ flow-api-pass 1))
+      (begin
+        (set! flow-api-fail (+ flow-api-fail 1))
+        (append! flow-api-fails {:name name :expected expected :actual actual})))))
+
+(define flow-a (fn (src) (flow-run src)))
+
+;; ── flow/status ─────────────────────────────────────────────────
+(flow-api-test "status: unknown id" (flow-a "(flow/status 999)") "unknown")
+(flow-api-test
+  "status: suspended flow"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/status id)")
+  "suspended")
+(flow-api-test
+  "status: completed flow"
+  (flow-a
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) v))) (define id (car (cdr (flow/start w 0)))) (flow/resume id 5) (flow/status id)")
+  "done")
+(flow-api-test
+  "status: cancelled flow"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/cancel id) (flow/status id)")
+  "cancelled")
+
+;; ── flow/result ─────────────────────────────────────────────────
+(flow-api-test
+  "result: returns the value of a completed flow"
+  (flow-a
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) (list (quote got) v)))) (define id (car (cdr (flow/start w 0)))) (flow/resume id 9) (flow/result id)")
+  (list "got" 9))
+(flow-api-test
+  "result: a still-suspended flow has no result"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/result id)")
+  (list "flow-error" "not-done"))
+(flow-api-test
+  "result: unknown id errors"
+  (flow-a "(flow/result 999)")
+  (list "flow-error" "no-such-flow"))
+
+;; ── flow/list ───────────────────────────────────────────────────
+(flow-api-test "list: empty store" (flow-a "(flow/list)") (list))
+(flow-api-test
+  "list: reports id + status for each flow (newest first)"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote q)))) (flow/start w 0) (flow/start (lambda (x) (* x 2)) 5) (flow/list)")
+  (list (list 2 "done") (list 1 "suspended")))
+
+;; ── flow/pending ────────────────────────────────────────────────
+(flow-api-test
+  "pending: lists suspended flows with their waiting tag"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote review)))) (flow/start w 0) (flow/pending)")
+  (list (list 1 "review")))
+(flow-api-test
+  "pending: excludes completed and cancelled flows"
+  (flow-a
+    "(defflow w (lambda (x) (suspend (quote q)))) (defflow v (sequence (lambda (x) (suspend (quote r))) (lambda (y) y))) (define i1 (car (cdr (flow/start w 0)))) (define i2 (car (cdr (flow/start v 0)))) (define i3 (car (cdr (flow/start w 0)))) (flow/resume i2 1) (flow/cancel i3) (flow/pending)")
+  (list (list 1 "q")))
+(flow-api-test
+  "pending: operator can drain all pending flows"
+  (flow-a
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) (* v 10)))) (flow/start w 0) (flow/start w 0) (define ps (flow/pending)) (flow/resume (car (car ps)) 1) (flow/resume (car (car (cdr ps))) 2) (flow/list)")
+  (list (list 1 "done") (list 2 "done")))
+
+(define flow-api-tests-run! (fn () {:total (+ flow-api-pass flow-api-fail) :passed flow-api-pass :failed flow-api-fail :fails flow-api-fails}))

lib/flow/tests/basic.sx

@@ -0,0 +1,121 @@
+;; lib/flow/tests/basic.sx — Phase 1: declarative DAG + sequential execution.
+
+(define flow-basic-pass 0)
+(define flow-basic-fail 0)
+(define flow-basic-fails (list))
+
+(define
+  flow-basic-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-basic-pass (+ flow-basic-pass 1))
+      (begin
+        (set! flow-basic-fail (+ flow-basic-fail 1))
+        (append! flow-basic-fails {:name name :expected expected :actual actual})))))
+
+;; Run a Scheme flow-program string and return its final value.
+(define flow-b (fn (src) (flow-run src)))
+
+;; Scheme strings are boxed as {:scm-string "..."}; unwrap to a host string.
+(define flow-bs (fn (src) (get (flow-run src) :scm-string)))
+
+;; ── single node ─────────────────────────────────────────────────
+(flow-basic-test
+  "node: identity passes input through"
+  (flow-b "(flow/start flow-id 7)")
+  7)
+(flow-basic-test
+  "node: const ignores input"
+  (flow-b "(flow/start (flow-const 99) 1)")
+  99)
+(flow-basic-test
+  "node: bare lambda is a node"
+  (flow-b "(flow/start (lambda (x) (* x x)) 6)")
+  36)
+
+;; ── linear sequence ─────────────────────────────────────────────
+(flow-basic-test
+  "sequence: empty is identity"
+  (flow-b "(flow/start (sequence) 42)")
+  42)
+(flow-basic-test
+  "sequence: single child"
+  (flow-b "(flow/start (sequence (lambda (x) (+ x 1))) 41)")
+  42)
+(flow-basic-test
+  "sequence: two children thread"
+  (flow-b
+    "(flow/start (sequence (lambda (x) (+ x 1)) (lambda (x) (* x 10))) 4)")
+  50)
+(flow-basic-test
+  "sequence: three children thread"
+  (flow-b
+    "(flow/start (sequence (lambda (x) (+ x 1)) (lambda (x) (* x 2)) (lambda (x) (- x 3))) 5)")
+  9)
+
+;; ── data flow between nodes ─────────────────────────────────────
+(flow-basic-test
+  "data flow: string accumulation"
+  (flow-bs
+    "(flow/start (sequence (lambda (s) (string-append s \"-a\")) (lambda (s) (string-append s \"-b\"))) \"x\")")
+  "x-a-b")
+(flow-basic-test
+  "data flow: list build"
+  (flow-b
+    "(flow/start (sequence (lambda (x) (cons x (list))) (lambda (xs) (cons 0 xs))) 7)")
+  (list 0 7))
+
+;; ── defflow ─────────────────────────────────────────────────────
+(flow-basic-test
+  "defflow: names a flow"
+  (flow-b
+    "(defflow inc2 (sequence (lambda (x) (+ x 1)) (lambda (x) (+ x 1)))) (flow/start inc2 40)")
+  42)
+(flow-basic-test
+  "defflow: reusable"
+  (flow-b
+    "(defflow dbl (lambda (x) (* x 2))) (+ (flow/start dbl 3) (flow/start dbl 10))")
+  26)
+
+;; ── parallel (sequential semantics, join into list) ─────────────
+(flow-basic-test
+  "parallel: fans input to all branches"
+  (flow-b
+    "(flow/start (parallel (lambda (x) (+ x 1)) (lambda (x) (* x 2)) (lambda (x) (- x 3))) 10)")
+  (list 11 20 7))
+(flow-basic-test
+  "parallel: empty joins to empty list"
+  (flow-b "(flow/start (parallel) 5)")
+  (list))
+(flow-basic-test
+  "parallel: single branch"
+  (flow-b "(flow/start (parallel (lambda (x) (* x x))) 9)")
+  (list 81))
+
+;; ── nested composition ──────────────────────────────────────────
+(flow-basic-test
+  "nested: sequence of sequences"
+  (flow-b
+    "(flow/start (sequence (sequence (lambda (x) (+ x 1)) (lambda (x) (+ x 1))) (sequence (lambda (x) (* x 3)))) 0)")
+  6)
+(flow-basic-test
+  "nested: parallel inside sequence, join then reduce"
+  (flow-b
+    "(flow/start (sequence (parallel (lambda (x) (+ x 1)) (lambda (x) (* x 2))) (lambda (xs) (apply + xs))) 10)")
+  31)
+(flow-basic-test
+  "nested: sequence inside parallel branch"
+  (flow-b
+    "(flow/start (parallel (sequence (lambda (x) (+ x 1)) (lambda (x) (* x 2))) (lambda (x) x)) 5)")
+  (list 12 5))
+
+;; ── publish-shaped flow (the architecture sketch) ───────────────
+(flow-basic-test
+  "publish: write -> (review | spell) -> join lengths"
+  (flow-b
+    "(defflow publish (sequence (lambda (draft) (string-append draft \"!\")) (parallel (lambda (c) (string-length c)) (lambda (c) (string-length (string-append c \"?\")))))) (flow/start publish \"hi\")")
+  (list 3 4))
+
+(define flow-basic-tests-run! (fn () {:total (+ flow-basic-pass flow-basic-fail) :passed flow-basic-pass :failed flow-basic-fail :fails flow-basic-fails}))

lib/flow/tests/combinators.sx

@@ -0,0 +1,108 @@
+;; lib/flow/tests/combinators.sx — Phase 5: combinator library (tap, recover, map-flow, iteration).
+
+(define flow-cmb-pass 0)
+(define flow-cmb-fail 0)
+(define flow-cmb-fails (list))
+
+(define
+  flow-cmb-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-cmb-pass (+ flow-cmb-pass 1))
+      (begin
+        (set! flow-cmb-fail (+ flow-cmb-fail 1))
+        (append! flow-cmb-fails {:name name :expected expected :actual actual})))))
+
+(define flow-m (fn (src) (flow-run src)))
+
+;; ── tap (side-effecting pass-through) ───────────────────────────
+(flow-cmb-test
+  "tap: returns input unchanged"
+  (flow-m "(flow/start (tap (lambda (x) (* x 999))) 7)")
+  7)
+(flow-cmb-test
+  "tap: runs the side effect"
+  (flow-m
+    "(define seen 0) (flow/start (tap (lambda (x) (set! seen x))) 42) seen")
+  42)
+(flow-cmb-test
+  "tap: value flows on while the effect observes it"
+  (flow-m
+    "(define log 0) (flow/start (sequence (lambda (x) (+ x 1)) (tap (lambda (x) (set! log x))) (lambda (x) (* x 2))) 10) (list log (flow/result 1))")
+  (list 11 22))
+
+;; ── recover (fail-value counterpart of try-catch) ───────────────
+(flow-cmb-test
+  "recover: passes a non-fail value through"
+  (flow-m "(flow/start (recover (lambda (x) (* x 2)) (lambda (r) -1)) 5)")
+  10)
+(flow-cmb-test
+  "recover: handles a fail value via the reason"
+  (flow-m
+    "(flow/start (recover (lambda (x) (fail (quote too-small))) (lambda (r) (list (quote recovered) r))) 1)")
+  (list "recovered" "too-small"))
+(flow-cmb-test
+  "recover: handler can supply a default value"
+  (flow-m
+    "(flow/start (sequence (recover (lambda (x) (if (> x 0) x (fail (quote neg))) ) (flow-const 0)) (lambda (x) (* x 10))) -3)")
+  0)
+(flow-cmb-test
+  "recover: does not catch raised exceptions (those are try-catch's job)"
+  (flow-m
+    "(flow/start (try-catch (recover (lambda (x) (raise (quote boom))) (flow-const 0)) (lambda (e) e)) 1)")
+  "boom")
+
+;; ── map-flow (run a node over a list, join) ─────────────────────
+(flow-cmb-test
+  "map-flow: applies the node to each item"
+  (flow-m "(flow/start (map-flow (lambda (x) (* x x))) (list 1 2 3 4))")
+  (list 1 4 9 16))
+(flow-cmb-test
+  "map-flow: empty list joins to empty"
+  (flow-m "(flow/start (map-flow (lambda (x) (+ x 1))) (list))")
+  (list))
+(flow-cmb-test
+  "map-flow: each item runs an independent sub-flow"
+  (flow-m
+    "(flow/start (map-flow (sequence (lambda (x) (+ x 1)) (lambda (x) (* x 2)))) (list 0 4 9))")
+  (list 2 10 20))
+(flow-cmb-test
+  "map-flow: composes — fan over a list then reduce the join"
+  (flow-m
+    "(flow/start (sequence (map-flow (lambda (x) (* x 10))) (lambda (xs) (apply + xs))) (list 1 2 3))")
+  60)
+
+;; ── flow-while / flow-until (bounded iteration) ─────────────────
+(flow-cmb-test
+  "flow-while: iterates while the predicate holds"
+  (flow-m
+    "(flow/start (flow-while (lambda (x) (< x 10)) (lambda (x) (+ x 1)) 100) 0)")
+  10)
+(flow-cmb-test
+  "flow-while: a false predicate leaves input unchanged"
+  (flow-m
+    "(flow/start (flow-while (lambda (x) (< x 0)) (lambda (x) (+ x 1)) 100) 5)")
+  5)
+(flow-cmb-test
+  "flow-while: respects the max-iteration bound"
+  (flow-m "(flow/start (flow-while (lambda (x) #t) (lambda (x) (+ x 1)) 3) 0)")
+  3)
+(flow-cmb-test
+  "flow-while: doubles until past a threshold"
+  (flow-m
+    "(flow/start (flow-while (lambda (x) (< x 50)) (lambda (x) (* x 2)) 100) 3)")
+  96)
+(flow-cmb-test
+  "flow-until: iterates until the predicate becomes true"
+  (flow-m
+    "(flow/start (flow-until (lambda (x) (>= x 10)) (lambda (x) (+ x 3)) 100) 0)")
+  12)
+(flow-cmb-test
+  "flow-until: composes inside a sequence"
+  (flow-m
+    "(flow/start (sequence (flow-until (lambda (x) (> x 100)) (lambda (x) (* x 3)) 100) (lambda (x) (- x 100))) 5)")
+  35)
+
+(define flow-cmb-tests-run! (fn () {:total (+ flow-cmb-pass flow-cmb-fail) :passed flow-cmb-pass :failed flow-cmb-fail :fails flow-cmb-fails}))

lib/flow/tests/control.sx

@@ -0,0 +1,179 @@
+;; lib/flow/tests/control.sx — Phase 2: control flow + error handling.
+
+(define flow-ctl-pass 0)
+(define flow-ctl-fail 0)
+(define flow-ctl-fails (list))
+
+(define
+  flow-ctl-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-ctl-pass (+ flow-ctl-pass 1))
+      (begin
+        (set! flow-ctl-fail (+ flow-ctl-fail 1))
+        (append! flow-ctl-fails {:name name :expected expected :actual actual})))))
+
+(define flow-c (fn (src) (flow-run src)))
+(define flow-cs (fn (src) (get (flow-run src) :scm-string)))
+
+;; ── branch ──────────────────────────────────────────────────────
+(flow-ctl-test
+  "branch: true selects then"
+  (flow-c
+    "(flow/start (branch (lambda (x) (> x 0)) (lambda (x) (* x 100)) (lambda (x) (- 0 x))) 5)")
+  500)
+(flow-ctl-test
+  "branch: false selects else"
+  (flow-c
+    "(flow/start (branch (lambda (x) (> x 0)) (lambda (x) (* x 100)) (lambda (x) (- 0 x))) -3)")
+  3)
+(flow-ctl-test
+  "branch: predicate sees the threaded input"
+  (flow-c
+    "(flow/start (sequence (lambda (x) (+ x 1)) (branch (lambda (x) (> x 3)) (flow-const 100) (flow-const 0))) 3)")
+  100)
+(flow-ctl-test
+  "branch: branches are full nodes (sequence inside)"
+  (flow-c
+    "(flow/start (branch (lambda (x) (< x 10)) (sequence (lambda (x) (+ x 1)) (lambda (x) (* x 2))) (flow-const 0)) 4)")
+  10)
+(flow-ctl-test
+  "branch: nested branch (3-way sign)"
+  (flow-c
+    "(defflow sign (branch (lambda (x) (> x 0)) (flow-const 1) (branch (lambda (x) (< x 0)) (flow-const -1) (flow-const 0)))) (list (flow/start sign 7) (flow/start sign -7) (flow/start sign 0))")
+  (list 1 -1 0))
+(flow-ctl-test
+  "branch: publish-shaped approval gate"
+  (flow-cs
+    "(defflow publish (branch (lambda (post) (>= (string-length post) 3)) (lambda (post) (string-append post \" [published]\")) (lambda (post) (string-append post \" [rejected]\")))) (flow/start publish \"ok\")")
+  "ok [rejected]")
+
+;; ── error model — explicit (fail reason) values ─────────────────
+(flow-ctl-test
+  "fail: failed? is true for a failure value"
+  (flow-c "(failed? (fail 404))")
+  true)
+(flow-ctl-test
+  "fail: fail-reason extracts the reason"
+  (flow-c "(fail-reason (fail 404))")
+  404)
+(flow-ctl-test
+  "fail: failed? is false for a plain value"
+  (flow-c "(failed? 7)")
+  false)
+(flow-ctl-test
+  "fail: failed? is false for an ordinary list"
+  (flow-c "(failed? (list 1 2 3))")
+  false)
+(flow-ctl-test
+  "fail: a node may emit a failure as data"
+  (flow-c
+    "(defflow validate (lambda (s) (if (>= (string-length s) 3) s (fail (quote too-short))))) (failed? (flow/start validate \"hi\"))")
+  true)
+(flow-ctl-test
+  "fail: failure flows downstream, branch recovers"
+  (flow-c
+    "(defflow guarded (sequence (lambda (s) (if (>= (string-length s) 3) (string-length s) (fail (quote too-short)))) (branch failed? (lambda (f) (list (quote recovered) (fail-reason f))) (lambda (n) (list (quote ok) n))))) (flow/start guarded \"hi\")")
+  (list "recovered" "too-short"))
+
+;; ── try-catch — reify raised exceptions ─────────────────────────
+(flow-ctl-test
+  "try-catch: no exception returns node result"
+  (flow-c "(flow/start (try-catch (lambda (x) (* x 2)) (lambda (e) -1)) 5)")
+  10)
+(flow-ctl-test
+  "try-catch: handler runs on raise"
+  (flow-c
+    "(flow/start (try-catch (lambda (x) (raise (quote boom))) (flow-const 99)) 1)")
+  99)
+(flow-ctl-test
+  "try-catch: handler receives the reified error"
+  (flow-c "(flow/start (try-catch (lambda (x) (raise 42)) (lambda (e) e)) 0)")
+  42)
+(flow-ctl-test
+  "try-catch: catches exception from deep inside a sequence"
+  (flow-c
+    "(flow/start (try-catch (sequence (lambda (x) (+ x 1)) (lambda (x) (raise (quote deep)))) (flow-const -99)) 5)")
+  -99)
+(flow-ctl-test
+  "try-catch: handler may convert to a failure value"
+  (flow-c
+    "(failed? (flow/start (try-catch (lambda (x) (raise (quote bad))) (lambda (e) (fail e))) 0))")
+  true)
+(flow-ctl-test
+  "try-catch: composes — recover then continue"
+  (flow-c
+    "(flow/start (sequence (try-catch (lambda (x) (raise (quote x))) (flow-const 10)) (lambda (n) (* n 5))) 0)")
+  50)
+
+;; ── retry — re-run on raised exceptions ─────────────────────────
+(flow-ctl-test
+  "retry: succeeds after transient failures"
+  (flow-c
+    "(define ctr 0) (defflow flaky (lambda (x) (set! ctr (+ ctr 1)) (if (< ctr 3) (raise (quote nope)) (* x 10)))) (list (flow/start (retry 5 flaky) 7) ctr)")
+  (list 70 3))
+(flow-ctl-test
+  "retry: exhausted re-raises (caught by try-catch)"
+  (flow-c
+    "(flow/start (try-catch (retry 2 (lambda (x) (raise (quote always)))) (flow-const (quote gaveup))) 0)")
+  "gaveup")
+(flow-ctl-test
+  "retry: n=1 means a single attempt"
+  (flow-c
+    "(define ctr 0) (flow/start (try-catch (retry 1 (lambda (x) (set! ctr (+ ctr 1)) (raise (quote bad)))) (lambda (e) ctr)) 0)")
+  1)
+(flow-ctl-test
+  "retry: success on first attempt does not re-run"
+  (flow-c
+    "(define ctr 0) (flow/start (sequence (retry 5 (lambda (x) (set! ctr (+ ctr 1)) (* x 2))) (lambda (n) ctr)) 21)")
+  1)
+(flow-ctl-test
+  "retry: does not retry explicit failure values"
+  (flow-c
+    "(define ctr 0) (failed? (flow/start (retry 5 (lambda (x) (set! ctr (+ ctr 1)) (fail (quote bad)))) 0))")
+  true)
+(flow-ctl-test
+  "retry: failure-value path runs node exactly once"
+  (flow-c
+    "(define ctr 0) (flow/start (sequence (retry 5 (lambda (x) (set! ctr (+ ctr 1)) (fail (quote bad)))) (lambda (f) ctr)) 0)")
+  1)
+
+;; ── timeout — cooperative step budget ───────────────────────────
+(flow-ctl-test
+  "timeout: work within budget completes"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 99 (begin (tick) (cd (- n 1))))) (flow/start (try-catch (timeout 10 (lambda (x) (cd x))) (flow-const (quote timed-out))) 5)")
+  99)
+(flow-ctl-test
+  "timeout: work exceeding budget raises flow-timeout"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 99 (begin (tick) (cd (- n 1))))) (flow/start (try-catch (timeout 10 (lambda (x) (cd x))) (flow-const (quote timed-out))) 20)")
+  "timed-out")
+(flow-ctl-test
+  "timeout: exact budget boundary completes"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 99 (begin (tick) (cd (- n 1))))) (flow/start (try-catch (timeout 5 (lambda (x) (cd x))) (flow-const (quote timed-out))) 5)")
+  99)
+(flow-ctl-test
+  "timeout: one tick over the budget raises"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 99 (begin (tick) (cd (- n 1))))) (flow/start (try-catch (timeout 5 (lambda (x) (cd x))) (flow-const (quote timed-out))) 6)")
+  "timed-out")
+(flow-ctl-test
+  "timeout: the raised error is identifiable"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 99 (begin (tick) (cd (- n 1))))) (flow/start (try-catch (timeout 2 (lambda (x) (cd x))) (lambda (e) e)) 9)")
+  "flow-timeout")
+(flow-ctl-test
+  "timeout: a node that never ticks is unbounded"
+  (flow-c "(flow/start (timeout 0 (lambda (x) (* x 2))) 5)")
+  10)
+(flow-ctl-test
+  "timeout: budget is restored across sequential timeouts"
+  (flow-c
+    "(define (cd n) (if (<= n 0) 1 (begin (tick) (cd (- n 1))))) (flow/start (sequence (timeout 4 (lambda (x) (cd x))) (timeout 4 (lambda (x) (cd 3))) (lambda (x) (begin (tick) (+ x 100)))) 3)")
+  101)
+
+(define flow-ctl-tests-run! (fn () {:total (+ flow-ctl-pass flow-ctl-fail) :passed flow-ctl-pass :failed flow-ctl-fail :fails flow-ctl-fails}))

lib/flow/tests/distributed.sx

@@ -0,0 +1,120 @@
+;; lib/flow/tests/distributed.sx — Phase 4: distributed nodes via fed-sx (mocked).
+
+(define flow-dist-pass 0)
+(define flow-dist-fail 0)
+(define flow-dist-fails (list))
+
+(define
+  flow-dist-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-dist-pass (+ flow-dist-pass 1))
+      (begin
+        (set! flow-dist-fail (+ flow-dist-fail 1))
+        (append! flow-dist-fails {:name name :expected expected :actual actual})))))
+
+(define flow-d (fn (src) (flow-run src)))
+
+;; ── remote-node ─────────────────────────────────────────────────
+(flow-dist-test
+  "remote: a node executes on a peer"
+  (flow-d
+    "(flow-peer-register! (quote edge) (list (list (quote double) (lambda (x) (* x 2))))) (flow/start (remote-node (quote edge) (quote double)) 21)")
+  42)
+(flow-dist-test
+  "remote: remote nodes compose in a sequence"
+  (flow-d
+    "(flow-peer-register! (quote edge) (list (list (quote inc) (lambda (x) (+ x 1))) (list (quote double) (lambda (x) (* x 2))))) (flow/start (sequence (remote-node (quote edge) (quote inc)) (remote-node (quote edge) (quote double))) 4)")
+  10)
+(flow-dist-test
+  "remote: a remote node mixes with local nodes"
+  (flow-d
+    "(flow-peer-register! (quote edge) (list (list (quote double) (lambda (x) (* x 2))))) (flow/start (sequence (lambda (x) (+ x 5)) (remote-node (quote edge) (quote double)) (lambda (x) (- x 1))) 10)")
+  29)
+(flow-dist-test
+  "remote: unreachable peer raises flow-remote-unreachable"
+  (flow-d
+    "(flow/start (try-catch (remote-node (quote ghost) (quote double)) (lambda (e) e)) 1)")
+  "flow-remote-unreachable")
+(flow-dist-test
+  "remote: unknown function on a peer raises flow-remote-no-fn"
+  (flow-d
+    "(flow-peer-register! (quote edge) (list (list (quote double) (lambda (x) (* x 2))))) (flow/start (try-catch (remote-node (quote edge) (quote missing)) (lambda (e) e)) 1)")
+  "flow-remote-no-fn")
+(flow-dist-test
+  "remote: a remote node can suspend the flow (peer returns control)"
+  (flow-d
+    "(flow-peer-register! (quote edge) (list (list (quote review) (lambda (x) x)))) (flow/start (sequence (remote-node (quote edge) (quote review)) (lambda (x) (suspend (quote human))) (lambda (v) (list (quote published) v))) 7)")
+  (list "flow-suspended" 1 "human"))
+(flow-dist-test
+  "remote: a transient remote failure is recoverable with retry"
+  (flow-d
+    "(define hits 0) (flow-peer-register! (quote edge) (list (list (quote flaky) (lambda (x) (begin (set! hits (+ hits 1)) (if (< hits 2) (raise (quote down)) (* x 3))))))) (list (flow/start (retry 3 (remote-node (quote edge) (quote flaky))) 7) hits)")
+  (list 21 2))
+
+;; ── failover (retry on a different peer, fall through to local) ──
+(flow-dist-test
+  "failover: first reachable peer serves the request"
+  (flow-d
+    "(flow-peer-register! (quote p2) (list (list (quote f) (lambda (x) (+ x 100))))) (flow/start (remote-failover (list (quote p2) (quote down)) (quote f) (flow-const (quote local))) 5)")
+  105)
+(flow-dist-test
+  "failover: skips an unreachable peer to the next one"
+  (flow-d
+    "(flow-peer-register! (quote p2) (list (list (quote f) (lambda (x) (+ x 100))))) (flow/start (remote-failover (list (quote down) (quote p2)) (quote f) (flow-const (quote local))) 5)")
+  105)
+(flow-dist-test
+  "failover: skips a peer whose function raises"
+  (flow-d
+    "(flow-peer-register! (quote bad) (list (list (quote f) (lambda (x) (raise (quote boom)))))) (flow-peer-register! (quote good) (list (list (quote f) (lambda (x) (* x 10))))) (flow/start (remote-failover (list (quote bad) (quote good)) (quote f) (flow-const 0)) 4)")
+  40)
+(flow-dist-test
+  "failover: all peers fail, the local fallback runs"
+  (flow-d
+    "(flow/start (remote-failover (list (quote down1) (quote down2)) (quote f) (lambda (x) (* x -1))) 9)")
+  -9)
+(flow-dist-test
+  "failover: threads the input through to the chosen peer"
+  (flow-d
+    "(flow-peer-register! (quote p) (list (list (quote f) (lambda (x) (list (quote got) x))))) (flow/start (sequence (lambda (x) (+ x 1)) (remote-failover (list (quote p)) (quote f) (flow-const 0))) 41)")
+  (list "got" 42))
+(flow-dist-test
+  "failover: composes inside a larger sequence"
+  (flow-d
+    "(flow-peer-register! (quote p) (list (list (quote f) (lambda (x) (* x 2))))) (flow/start (sequence (remote-failover (list (quote down) (quote p)) (quote f) (flow-const 1)) (lambda (x) (+ x 3))) 5)")
+  13)
+
+;; ── replication + handoff ───────────────────────────────────────
+(flow-dist-test
+  "replicate: a peer holds the exported store"
+  (flow-d
+    "(defflow w (lambda (x) (suspend (quote q)))) (flow/start w 10) (flow-replicate-to (quote peerB)) (if (flow-replica-get (quote peerB)) (quote replicated) (quote missing))")
+  "replicated")
+(flow-dist-test
+  "handoff: a peer resumes a flow after the local instance dies"
+  (flow-d
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) (list (quote done) v)))) (define id (car (cdr (flow/start w 10)))) (flow-replicate-to (quote peerB)) (set! flow-store (list)) (flow-restore-from (quote peerB)) (flow/resume id 55)")
+  (list "done" 55))
+(flow-dist-test
+  "handoff: restored peer reports the flow as resumable"
+  (flow-d
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 10)))) (flow-replicate-to (quote peerB)) (set! flow-store (list)) (flow-restore-from (quote peerB)) (flow-resumable-ids)")
+  (list 1))
+(flow-dist-test
+  "handoff: without restore the dead instance has lost the flow"
+  (flow-d
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 10)))) (flow-replicate-to (quote peerB)) (set! flow-store (list)) (flow/resume id 1)")
+  (list "flow-error" "no-such-flow"))
+(flow-dist-test
+  "restore: from an unknown peer yields false"
+  (flow-d "(flow-restore-from (quote nowhere))")
+  false)
+(flow-dist-test
+  "handoff: replication preserves the replay log across the move"
+  (flow-d
+    "(defflow two (sequence (lambda (x) (suspend (quote a))) (lambda (x) (suspend (quote b))) (lambda (x) (list x)))) (define id (car (cdr (flow/start two 0)))) (flow/resume id 11) (flow-replicate-to (quote peerB)) (set! flow-store (list)) (flow-restore-from (quote peerB)) (flow/resume id 22)")
+  (list 22))
+
+(define flow-dist-tests-run! (fn () {:total (+ flow-dist-pass flow-dist-fail) :passed flow-dist-pass :failed flow-dist-fail :fails flow-dist-fails}))

lib/flow/tests/host.sx

@@ -0,0 +1,106 @@
+;; lib/flow/tests/host.sx — Phase 8: host integration ABI (request/await/host-queue/driver).
+
+(define flow-hst-pass 0)
+(define flow-hst-fail 0)
+(define flow-hst-fails (list))
+
+(define
+  flow-hst-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-hst-pass (+ flow-hst-pass 1))
+      (begin
+        (set! flow-hst-fail (+ flow-hst-fail 1))
+        (append! flow-hst-fails {:name name :expected expected :actual actual})))))
+
+(define flow-hst (fn (src) (flow-run src)))
+
+;; ── request envelope ────────────────────────────────────────────
+(flow-hst-test
+  "request: suspends with a typed envelope"
+  (flow-hst
+    "(car (cdr (cdr (flow/start (lambda (x) (request (quote render) x)) 5))))")
+  (list "flow-request" "render" 5))
+(flow-hst-test
+  "request?: recognizes an envelope"
+  (flow-hst "(request? (list (quote flow-request) (quote human) 1))")
+  true)
+(flow-hst-test
+  "request?: a plain tag is not a request"
+  (flow-hst "(request? (list (quote review) 1))")
+  false)
+(flow-hst-test
+  "request-kind / request-payload: parse the envelope"
+  (flow-hst
+    "(define t (list (quote flow-request) (quote render) (list (quote recipe) 7))) (list (request-kind t) (request-payload t))")
+  (list "render" (list "recipe" 7)))
+
+;; ── named decision points ───────────────────────────────────────
+(flow-hst-test
+  "await-human: is a request of kind human"
+  (flow-hst
+    "(car (cdr (cdr (flow/start (lambda (x) (await-human x)) (quote approve?)))))")
+  (list "flow-request" "human" "approve?"))
+(flow-hst-test
+  "await-render: is a request of kind render"
+  (flow-hst
+    "(car (cdr (cdr (flow/start (lambda (x) (await-render x)) (quote recipe)))))")
+  (list "flow-request" "render" "recipe"))
+(flow-hst-test
+  "request: the host's resume value flows back into the flow"
+  (flow-hst
+    "(defflow f (sequence (lambda (x) (await-render x)) (lambda (art) (list (quote got) art)))) (define id (car (cdr (flow/start f 1)))) (flow/resume id (quote the-artifact))")
+  (list "got" "the-artifact"))
+
+;; ── host work queue ─────────────────────────────────────────────
+(flow-hst-test
+  "flow-host-requests: lists (id kind payload) for pending requests"
+  (flow-hst
+    "(flow/start (lambda (x) (await-render x)) 99) (flow-host-requests)")
+  (list (list 1 "render" 99)))
+(flow-hst-test
+  "flow-host-requests: excludes bare (non-request) suspends"
+  (flow-hst
+    "(defflow a (lambda (x) (await-render x))) (defflow b (lambda (x) (suspend (quote plain)))) (flow/start a 1) (flow/start b 2) (flow-host-requests)")
+  (list (list 1 "render" 1)))
+
+;; ── the art-dag-shaped host driver loop (manual resumes) ────────
+(flow-hst-test
+  "host driver: render then human-review then publish"
+  (flow-hst
+    "(defflow pipeline (sequence (lambda (recipe) (await-render recipe)) (lambda (art) (await-human (list (quote review) art))) (branch (lambda (d) (eq? d (quote approve))) (flow-const (quote published)) (flow-const (fail (quote rejected)))))) (define id (car (cdr (flow/start pipeline 99)))) (define r1 (flow-host-requests)) (flow/resume id (list (quote art) 99)) (define r2 (flow-host-requests)) (flow/resume id (quote approve)) (list r1 r2 (flow/status id) (flow/result id))")
+  (list
+    (list (list 1 "render" 99))
+    (list (list 1 "human" (list "review" (list "art" 99))))
+    "done"
+    "published"))
+(flow-hst-test
+  "host driver: rejection at the human gate yields a failure"
+  (flow-hst
+    "(defflow pipeline (sequence (lambda (recipe) (await-render recipe)) (lambda (art) (await-human (list (quote review) art))) (branch (lambda (d) (eq? d (quote approve))) (flow-const (quote published)) (flow-const (fail (quote rejected)))))) (define id (car (cdr (flow/start pipeline 1)))) (flow/resume id (quote artifact)) (failed? (flow/resume id (quote reject)))")
+  true)
+
+;; ── reference driver: host supplies only a dispatch fn ──────────
+(flow-hst-test
+  "flow-drive-host: one tick services every pending request"
+  (flow-hst
+    "(flow/start (lambda (x) (await-render x)) 5) (define n (flow-drive-host (lambda (k p) (list (quote done) p)))) (list n (flow/status 1) (flow/result 1))")
+  (list 1 "done" (list "done" 5)))
+(flow-hst-test
+  "flow-run-host: drives a render -> human pipeline to completion"
+  (flow-hst
+    "(defflow pipeline (sequence (lambda (recipe) (await-render recipe)) (lambda (art) (await-human (list (quote review) art))) (branch (lambda (d) (eq? d (quote approve))) (flow-const (quote published)) (flow-const (fail (quote rejected)))))) (define id (car (cdr (flow/start pipeline 99)))) (define serviced (flow-run-host (lambda (kind payload) (if (eq? kind (quote render)) (list (quote art) payload) (quote approve))) 10)) (list serviced (flow/status id) (flow/result id))")
+  (list 2 "done" "published"))
+(flow-hst-test
+  "flow-run-host: returns 0 when nothing is pending"
+  (flow-hst "(flow-run-host (lambda (k p) p) 5)")
+  0)
+(flow-hst-test
+  "flow-run-host: respects the maxticks bound"
+  (flow-hst
+    "(defflow pipe2 (sequence (lambda (r) (await-render r)) (lambda (a) (await-human a)) (lambda (d) d))) (define id (car (cdr (flow/start pipe2 1)))) (define serviced (flow-run-host (lambda (k p) p) 1)) (list serviced (flow/status id))")
+  (list 1 "suspended"))
+
+(define flow-hst-tests-run! (fn () {:total (+ flow-hst-pass flow-hst-fail) :passed flow-hst-pass :failed flow-hst-fail :fails flow-hst-fails}))

lib/flow/tests/hygiene.sx

@@ -0,0 +1,67 @@
+;; lib/flow/tests/hygiene.sx — Phase 5: store hygiene (flow/gc, flow/forget).
+
+(define flow-hyg-pass 0)
+(define flow-hyg-fail 0)
+(define flow-hyg-fails (list))
+
+(define
+  flow-hyg-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-hyg-pass (+ flow-hyg-pass 1))
+      (begin
+        (set! flow-hyg-fail (+ flow-hyg-fail 1))
+        (append! flow-hyg-fails {:name name :expected expected :actual actual})))))
+
+(define flow-h (fn (src) (flow-run src)))
+
+;; ── flow/gc ─────────────────────────────────────────────────────
+(flow-hyg-test
+  "gc: empty store removes nothing"
+  (flow-h "(flow/gc)")
+  0)
+(flow-hyg-test
+  "gc: removes a done flow, keeps a suspended one"
+  (flow-h
+    "(defflow w (lambda (x) (suspend (quote q)))) (flow/start w 0) (flow/start (lambda (x) x) 5) (define removed (flow/gc)) (list removed (flow/list))")
+  (list 1 (list (list 1 "suspended"))))
+(flow-hyg-test
+  "gc: removes a cancelled flow"
+  (flow-h
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/cancel id) (flow/gc)")
+  1)
+(flow-hyg-test
+  "gc: a kept suspended flow is still resumable"
+  (flow-h
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) (* v 2)))) (define id (car (cdr (flow/start w 0)))) (flow/start (lambda (x) x) 1) (flow/gc) (flow/resume id 21)")
+  42)
+(flow-hyg-test
+  "gc: counts every terminal flow it drops"
+  (flow-h
+    "(flow/start (lambda (x) x) 1) (flow/start (lambda (x) x) 2) (defflow w (lambda (x) (suspend (quote q)))) (flow/start w 0) (flow/gc)")
+  2)
+
+;; ── flow/forget ─────────────────────────────────────────────────
+(flow-hyg-test
+  "forget: drops a completed flow"
+  (flow-h
+    "(defflow w (sequence (lambda (x) (suspend (quote q))) (lambda (v) v))) (define id (car (cdr (flow/start w 0)))) (flow/resume id 7) (list (flow/forget id) (flow/status id))")
+  (list true "unknown"))
+(flow-hyg-test
+  "forget: refuses to drop a live (suspended) flow"
+  (flow-h
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (list (flow/forget id) (flow/status id))")
+  (list false "suspended"))
+(flow-hyg-test
+  "forget: drops a cancelled flow"
+  (flow-h
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/cancel id) (list (flow/forget id) (flow/status id))")
+  (list true "unknown"))
+(flow-hyg-test
+  "forget: unknown id yields false"
+  (flow-h "(flow/forget 999)")
+  false)
+
+(define flow-hyg-tests-run! (fn () {:total (+ flow-hyg-pass flow-hyg-fail) :passed flow-hyg-pass :failed flow-hyg-fail :fails flow-hyg-fails}))

lib/flow/tests/integration.sx

@@ -0,0 +1,115 @@
+;; lib/flow/tests/integration.sx — Phase 7: end-to-end flows composing every phase.
+
+(define flow-int-pass 0)
+(define flow-int-fail 0)
+(define flow-int-fails (list))
+
+(define
+  flow-int-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-int-pass (+ flow-int-pass 1))
+      (begin
+        (set! flow-int-fail (+ flow-int-fail 1))
+        (append! flow-int-fails {:name name :expected expected :actual actual})))))
+
+(define flow-i (fn (src) (flow-run src)))
+
+;; The order-processing flow, defined once per program via this prelude string:
+;;   validate amount (attempt: fail if <= 0)
+;;   -> suspend for payment confirmation (resume value = confirmed amount)
+;;   -> branch: confirmed>0 ? record on the ledger peer : declined failure
+(define
+  order-prelude
+  "(flow-peer-register! (quote ledger) (list (list (quote record) (lambda (amt) (list (quote recorded) amt)))))\n   (defflow order\n     (attempt\n       (lambda (amt) (if (> amt 0) amt (fail (quote invalid-amount))))\n       (lambda (amt) (suspend (quote await-payment)))\n       (branch (lambda (amt) (> amt 0))\n               (remote-node (quote ledger) (quote record))\n               (flow-const (fail (quote declined))))))")
+
+;; ── happy path through every phase ──────────────────────────────
+(flow-int-test
+  "order: validate -> suspend -> resume -> branch -> federate"
+  (flow-i
+    (str
+      order-prelude
+      "(define id (car (cdr (flow/start order 100)))) (flow/resume id 250)"))
+  (list "recorded" 250))
+(flow-int-test
+  "order: starting suspends awaiting payment"
+  (flow-i
+    (str
+      order-prelude
+      "(define s (flow/start order 100)) (list (car s) (car (cdr (cdr s))))"))
+  (list "flow-suspended" "await-payment"))
+(flow-int-test
+  "order: invalid amount fails up front and never suspends"
+  (flow-i
+    (str
+      order-prelude
+      "(define r (flow/start order -5)) (list (failed? r) (fail-reason r))"))
+  (list true "invalid-amount"))
+(flow-int-test
+  "order: a declined payment yields a failure value"
+  (flow-i
+    (str
+      order-prelude
+      "(define id (car (cdr (flow/start order 100)))) (failed? (flow/resume id 0))"))
+  true)
+
+;; ── crash recovery mid-flow ─────────────────────────────────────
+(flow-int-test
+  "order: survives a simulated crash between suspend and resume"
+  (flow-i
+    (str
+      order-prelude
+      "(define id (car (cdr (flow/start order 100)))) (define saved (flow-store-export)) (set! flow-store (list)) (flow-store-import! saved) (flow/resume id 250)"))
+  (list "recorded" 250))
+
+;; ── handoff to a peer mid-flow ──────────────────────────────────
+(flow-int-test
+  "order: hands off to a peer that resumes and completes"
+  (flow-i
+    (str
+      order-prelude
+      "(define id (car (cdr (flow/start order 100)))) (flow-replicate-to (quote nodeB)) (set! flow-store (list)) (flow-restore-from (quote nodeB)) (flow/resume id 250)"))
+  (list "recorded" 250))
+
+;; ── introspection during the flow's life ────────────────────────
+(flow-int-test
+  "order: pending shows what the flow awaits, then result after resume"
+  (flow-i
+    (str
+      order-prelude
+      "(define id (car (cdr (flow/start order 100)))) (define p (flow/pending)) (flow/resume id 250) (list p (flow/status id) (flow/result id))"))
+  (list
+    (list (list 1 "await-payment"))
+    "done"
+    (list "recorded" 250)))
+
+;; ── onboarding: two human steps + cancellation ──────────────────
+(define
+  onboard-prelude
+  "(defflow onboard\n     (sequence\n       (lambda (user) (+ user 1))\n       (lambda (x) (suspend (quote confirm-email)))\n       (lambda (x) (suspend (quote complete-profile)))\n       (lambda (x) (list (quote onboarded) x))))")
+
+(flow-int-test
+  "onboard: two suspends resume in order to completion"
+  (flow-i
+    (str
+      onboard-prelude
+      "(define id (car (cdr (flow/start onboard 0)))) (flow/resume id 7) (flow/resume id 9)"))
+  (list "onboarded" 9))
+(flow-int-test
+  "onboard: the second pending tag appears after the first resume"
+  (flow-i
+    (str
+      onboard-prelude
+      "(define id (car (cdr (flow/start onboard 0)))) (flow/resume id 7) (car (cdr (car (flow/pending))))"))
+  "complete-profile")
+(flow-int-test
+  "onboard: cancelling abandons the flow"
+  (flow-i
+    (str
+      onboard-prelude
+      "(define id (car (cdr (flow/start onboard 0)))) (flow/cancel id) (list (flow/status id) (car (flow/resume id 7)))"))
+  (list "cancelled" "flow-error"))
+
+(define flow-int-tests-run! (fn () {:total (+ flow-int-pass flow-int-fail) :passed flow-int-pass :failed flow-int-fail :fails flow-int-fails}))

lib/flow/tests/railway.sx

@@ -0,0 +1,73 @@
+;; lib/flow/tests/railway.sx — Phase 6: railway-oriented composition (attempt).
+
+(define flow-rail-pass 0)
+(define flow-rail-fail 0)
+(define flow-rail-fails (list))
+
+(define
+  flow-rail-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-rail-pass (+ flow-rail-pass 1))
+      (begin
+        (set! flow-rail-fail (+ flow-rail-fail 1))
+        (append! flow-rail-fails {:name name :expected expected :actual actual})))))
+
+(define flow-r (fn (src) (flow-run src)))
+
+;; ── attempt — short-circuit on the first (fail ...) ─────────────
+(flow-rail-test
+  "attempt: threads like sequence when nothing fails"
+  (flow-r
+    "(flow/start (attempt (lambda (x) (+ x 1)) (lambda (x) (* x 10))) 4)")
+  50)
+(flow-rail-test
+  "attempt: empty is identity"
+  (flow-r "(flow/start (attempt) 7)")
+  7)
+(flow-rail-test
+  "attempt: returns the first failure"
+  (flow-r
+    "(failed? (flow/start (attempt (lambda (x) (fail (quote bad))) (lambda (x) (* x 10))) 4))")
+  true)
+(flow-rail-test
+  "attempt: the failure carries its reason"
+  (flow-r
+    "(fail-reason (flow/start (attempt (lambda (x) x) (lambda (x) (fail (quote rejected)))) 4))")
+  "rejected")
+(flow-rail-test
+  "attempt: nodes after a failure do not run"
+  (flow-r
+    "(define ran 0) (flow/start (attempt (lambda (x) (fail (quote stop))) (lambda (x) (begin (set! ran (+ ran 1)) x))) 0) ran")
+  0)
+(flow-rail-test
+  "attempt: a failed input short-circuits immediately"
+  (flow-r
+    "(define ran 0) (fail-reason (flow/start (attempt (lambda (x) (begin (set! ran (+ ran 1)) x))) (fail (quote pre))))")
+  "pre")
+(flow-rail-test
+  "attempt: middle failure halts the chain"
+  (flow-r
+    "(define ran 0) (flow/start (attempt (lambda (x) (+ x 1)) (lambda (x) (fail (quote mid))) (lambda (x) (begin (set! ran (+ ran 1)) x))) 5) ran")
+  0)
+
+;; ── attempt + recover (rejoin the happy track) ──────────────────
+(flow-rail-test
+  "attempt + recover: recover turns a failure into a value"
+  (flow-r
+    "(flow/start (recover (attempt (lambda (x) (if (> x 0) x (fail (quote non-positive)))) (lambda (x) (* x 2))) (flow-const 0)) -5)")
+  0)
+(flow-rail-test
+  "attempt + recover: happy path passes recover through"
+  (flow-r
+    "(flow/start (recover (attempt (lambda (x) (if (> x 0) x (fail (quote non-positive)))) (lambda (x) (* x 2))) (flow-const 0)) 5)")
+  10)
+(flow-rail-test
+  "attempt: validation pipeline reports the failing stage"
+  (flow-r
+    "(defflow validate (attempt (lambda (s) (if (>= (string-length s) 3) s (fail (quote too-short)))) (lambda (s) (if (<= (string-length s) 8) s (fail (quote too-long)))) (lambda (s) (list (quote ok) (string-length s))))) (list (fail-reason (flow/start validate \"hi\")) (flow/start validate \"hello\"))")
+  (list "too-short" (list "ok" 5)))
+
+(define flow-rail-tests-run! (fn () {:total (+ flow-rail-pass flow-rail-fail) :passed flow-rail-pass :failed flow-rail-fail :fails flow-rail-fails}))

lib/flow/tests/recovery.sx

@@ -0,0 +1,71 @@
+;; lib/flow/tests/recovery.sx — Phase 3: crash recovery (store export/import + restart).
+;;
+;; "restart" is simulated within one program: (set! flow-store (list)) wipes the
+;; in-memory store (process death), while flow-registry persists as it would after
+;; reloading flow definitions. Recovery = import the exported (plain-data) store and
+;; resume; the flow proc is re-resolved by name.
+
+(define flow-rec-pass 0)
+(define flow-rec-fail 0)
+(define flow-rec-fails (list))
+
+(define
+  flow-rec-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-rec-pass (+ flow-rec-pass 1))
+      (begin
+        (set! flow-rec-fail (+ flow-rec-fail 1))
+        (append! flow-rec-fails {:name name :expected expected :actual actual})))))
+
+(define flow-r (fn (src) (flow-run src)))
+
+;; ── export / wipe / import ──────────────────────────────────────
+(flow-rec-test
+  "export nulls the live procedure"
+  (flow-r
+    "(defflow w (lambda (x) (suspend (quote await)))) (flow/start w 10) (car (cdr (car (cdr (car (flow-store-export))))))")
+  false)
+(flow-rec-test
+  "a wiped store loses the flow (process death)"
+  (flow-r
+    "(defflow w (lambda (x) (suspend (quote await)))) (define id (car (cdr (flow/start w 10)))) (set! flow-store (list)) (flow/resume id 1)")
+  (list "flow-error" "no-such-flow"))
+(flow-rec-test
+  "import restores a wiped store and resume completes"
+  (flow-r
+    "(defflow w (sequence (lambda (x) (suspend (quote await))) (lambda (c) (list (quote done) c)))) (define id (car (cdr (flow/start w 10)))) (define saved (flow-store-export)) (set! flow-store (list)) (flow-store-import! saved) (flow/resume id 777)")
+  (list "done" 777))
+
+;; ── resumable scan ──────────────────────────────────────────────
+(flow-rec-test
+  "resumable-ids lists the suspended flow after import"
+  (flow-r
+    "(defflow w (lambda (x) (suspend (quote await)))) (define id (car (cdr (flow/start w 10)))) (define saved (flow-store-export)) (set! flow-store (list)) (flow-store-import! saved) (flow-resumable-ids)")
+  (list 1))
+(flow-rec-test
+  "resumable-ids excludes completed flows"
+  (flow-r
+    "(defflow w (sequence (lambda (x) (suspend (quote await))) (lambda (c) c))) (define id (car (cdr (flow/start w 10)))) (flow/resume id 5) (flow-resumable-ids)")
+  (list))
+(flow-rec-test
+  "resumable-ids excludes cancelled flows after import"
+  (flow-r
+    "(defflow w (lambda (x) (suspend (quote await)))) (define id (car (cdr (flow/start w 10)))) (flow/cancel id) (define saved (flow-store-export)) (set! flow-store (list)) (flow-store-import! saved) (flow-resumable-ids)")
+  (list))
+
+;; ── restart at every step ───────────────────────────────────────
+(flow-rec-test
+  "two suspends survive a restart between each step"
+  (flow-r
+    "(defflow two (sequence (lambda (x) (suspend (quote a))) (lambda (x) (suspend (quote b))) (lambda (x) (list (quote end) x)))) (define id (car (cdr (flow/start two 0)))) (define s1 (flow-store-export)) (set! flow-store (list)) (flow-store-import! s1) (flow/resume id 100) (define s2 (flow-store-export)) (set! flow-store (list)) (flow-store-import! s2) (flow/resume id 200)")
+  (list "end" 200))
+(flow-rec-test
+  "import preserves the replay log (earlier value survives restart)"
+  (flow-r
+    "(defflow two (sequence (lambda (x) (suspend (quote a))) (lambda (x) (suspend (quote b))) (lambda (x) (list x)))) (define id (car (cdr (flow/start two 0)))) (flow/resume id 11) (define saved (flow-store-export)) (set! flow-store (list)) (flow-store-import! saved) (flow/resume id 22)")
+  (list 22))
+
+(define flow-rec-tests-run! (fn () {:total (+ flow-rec-pass flow-rec-fail) :passed flow-rec-pass :failed flow-rec-fail :fails flow-rec-fails}))

lib/flow/tests/suspend.sx

@@ -0,0 +1,114 @@
+;; lib/flow/tests/suspend.sx — Phase 3: suspend / resume / cancel (deterministic replay).
+
+(define flow-sus-pass 0)
+(define flow-sus-fail 0)
+(define flow-sus-fails (list))
+
+(define
+  flow-sus-test
+  (fn
+    (name actual expected)
+    (if
+      (= actual expected)
+      (set! flow-sus-pass (+ flow-sus-pass 1))
+      (begin
+        (set! flow-sus-fail (+ flow-sus-fail 1))
+        (append! flow-sus-fails {:name name :expected expected :actual actual})))))
+
+(define flow-s (fn (src) (flow-run src)))
+
+;; ── flow/start ──────────────────────────────────────────────────
+(flow-sus-test
+  "start: non-suspending flow returns the raw result"
+  (flow-s "(flow/start (lambda (x) (* x 2)) 5)")
+  10)
+(flow-sus-test
+  "start: a suspending flow returns a flow-suspended state"
+  (flow-s
+    "(defflow w (sequence (lambda (x) (+ x 1)) (lambda (g) (suspend (quote await))) (lambda (c) c))) (car (flow/start w 10))")
+  "flow-suspended")
+(flow-sus-test
+  "start: suspended state carries a numeric id"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote await)))) (car (cdr (flow/start w 10)))")
+  1)
+(flow-sus-test
+  "start: suspended state carries the suspend tag"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote await)))) (car (cdr (cdr (flow/start w 10))))")
+  "await")
+
+;; ── flow/resume ─────────────────────────────────────────────────
+(flow-sus-test
+  "resume: injects the value and completes"
+  (flow-s
+    "(defflow w (sequence (lambda (x) (+ x 1)) (lambda (g) (suspend (quote await))) (lambda (c) (list (quote done) c)))) (define s (flow/start w 10)) (flow/resume (car (cdr s)) 777)")
+  (list "done" 777))
+(flow-sus-test
+  "resume: injected value threads into the next node"
+  (flow-s
+    "(defflow w (sequence (lambda (x) (suspend (quote v))) (lambda (n) (* n 3)))) (define s (flow/start w 0)) (flow/resume (car (cdr s)) 14)")
+  42)
+(flow-sus-test
+  "resume: replays earlier suspends (recompute is deterministic)"
+  (flow-s
+    "(define runs 0) (defflow w (sequence (lambda (x) (begin (set! runs (+ runs 1)) (+ x 1))) (lambda (g) (suspend (quote await))) (lambda (c) c))) (define s (flow/start w 10)) (flow/resume (car (cdr s)) 99) runs")
+  2)
+
+;; ── multi-step suspension ───────────────────────────────────────
+(flow-sus-test
+  "multi: first resume suspends at the next tag"
+  (flow-s
+    "(defflow two (sequence (lambda (x) (suspend (quote a))) (lambda (x) (suspend (quote b))) (lambda (x) (list (quote end) x)))) (define s (flow/start two 0)) (define s2 (flow/resume (car (cdr s)) 100)) (car (cdr (cdr s2)))")
+  "b")
+(flow-sus-test
+  "multi: second resume completes with the latest value"
+  (flow-s
+    "(defflow two (sequence (lambda (x) (suspend (quote a))) (lambda (x) (suspend (quote b))) (lambda (x) (list (quote end) x)))) (define id (car (cdr (flow/start two 0)))) (flow/resume id 100) (flow/resume id 200)")
+  (list "end" 200))
+
+;; ── error / lifecycle guards ────────────────────────────────────
+(flow-sus-test
+  "resume: completed flow cannot be resumed again"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/resume id 1) (flow/resume id 2)")
+  (list "flow-error" "not-suspended"))
+(flow-sus-test
+  "resume: unknown id errors"
+  (flow-s "(flow/resume 999 1)")
+  (list "flow-error" "no-such-flow"))
+
+;; ── flow/cancel ─────────────────────────────────────────────────
+(flow-sus-test
+  "cancel: returns a flow-cancelled state"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/cancel id)")
+  (list "flow-cancelled" 1))
+(flow-sus-test
+  "cancel: a cancelled flow cannot be resumed (stale resume rejected)"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote q)))) (define id (car (cdr (flow/start w 0)))) (flow/cancel id) (flow/resume id 5)")
+  (list "flow-error" "not-suspended"))
+(flow-sus-test
+  "cancel: unknown id errors"
+  (flow-s "(flow/cancel 999)")
+  (list "flow-error" "no-such-flow"))
+
+;; ── composition ─────────────────────────────────────────────────
+(flow-sus-test
+  "suspend inside a branch arm"
+  (flow-s
+    "(defflow gate (branch (lambda (x) (> x 0)) (lambda (x) (suspend (quote approve))) (flow-const (quote rejected)))) (define s (flow/start gate 5)) (flow/resume (car (cdr s)) (quote approved))")
+  "approved")
+(flow-sus-test
+  "two independent runs get independent ids"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (quote q)))) (list (car (cdr (flow/start w 0))) (car (cdr (flow/start w 0))))")
+  (list 1 2))
+(flow-sus-test
+  "suspend reason may be a structured value"
+  (flow-s
+    "(defflow w (lambda (x) (suspend (list (quote needs) (quote approval))))) (car (cdr (cdr (flow/start w 0))))")
+  (list "needs" "approval"))
+
+(define flow-sus-tests-run! (fn () {:total (+ flow-sus-pass flow-sus-fail) :passed flow-sus-pass :failed flow-sus-fail :fails flow-sus-fails}))

lib/mod/activity.sx

@@ -1,40 +0,0 @@
-;; lib/mod/activity.sx — export decisions as ActivityPub-shaped events.
-;;
-;; The rose-ash platform propagates cross-domain effects as ActivityPub-shaped
-;; activities. A moderation decision maps to a moderation verb so the rest of the
-;; platform (and federated peers) can act on it: remove→Delete, ban→Block,
-;; hide/escalate→Flag, keep→no activity. The precise mod action is preserved in
-;; :action so a consumer can disambiguate (e.g. hide vs escalate, both Flag).
-
-(define
-  mod/action->verb
-  (fn
-    (action)
-    (cond
-      ((= action "remove") "Delete")
-      ((= action "ban") "Block")
-      ((= action "hide") "Flag")
-      ((= action "escalate") "Flag")
-      (true nil))))
-
-(define
-  mod/decision->activity
-  (fn
-    (d actor)
-    (let
-      ((verb (mod/action->verb (get d :action))))
-      (if (nil? verb) nil {:type verb :action (get d :action) :actor actor :summary (str "moderation/" (get d :action) " via " (get d :rule)) :object (get d :report-id) :rule (get d :rule)}))))
-
-;; map a batch of decisions to activities, dropping the no-op keeps
-(define
-  mod/decisions->activities
-  (fn
-    (decisions actor)
-    (reduce
-      (fn
-        (acc d)
-        (let
-          ((a (mod/decision->activity d actor)))
-          (if (nil? a) acc (append acc (list a)))))
-      (list)
-      decisions)))

lib/mod/api.sx

@@ -1,163 +0,0 @@
-;; lib/mod/api.sx — report registry + lifecycle façade + public entry points.
-;;
-;; mod/report files a report (assigning a sequential id) and opens a lifecycle
-;; case for it; mod/add-evidence accumulates evidence; mod/decide runs the engine
-;; and commits to the audit log. The lifecycle façade (mod/triage, mod/resolve,
-;; mod/review, mod/appeal, mod/finalize) drives the per-report case through its
-;; states, logging each committed decision to the audit trail.
-
-(define mod/*reports* (list))
-(define mod/*cases* (list))
-(define mod/*counter* 0)
-(define mod/*rules* mod/default-rules)
-
-(define
-  mod/reset!
-  (fn
-    ()
-    (begin
-      (set! mod/*reports* (list))
-      (set! mod/*cases* (list))
-      (set! mod/*counter* 0)
-      (mod/audit-reset!))))
-
-(define
-  mod/report
-  (fn
-    (by about reason)
-    (begin
-      (set! mod/*counter* (+ mod/*counter* 1))
-      (let
-        ((id (str "r" mod/*counter*)))
-        (let
-          ((r (mod/mk-report id by about reason)))
-          (begin
-            (append! mod/*reports* r)
-            (append! mod/*cases* {:id id :case (mod/mk-case r)})
-            r))))))
-
-(define
-  mod/get-report
-  (fn
-    (id)
-    (reduce
-      (fn (acc r) (if (= (mod/report-id r) id) r acc))
-      nil
-      mod/*reports*)))
-
-(define
-  mod/add-evidence
-  (fn
-    (id kind val)
-    (let
-      ((r (mod/get-report id)))
-      (if
-        (nil? r)
-        nil
-        (let
-          ((updated (mod/attach-evidence r (mod/mk-evidence kind val))))
-          (begin
-            (set!
-              mod/*reports*
-              (map
-                (fn (x) (if (= (mod/report-id x) id) updated x))
-                mod/*reports*))
-            updated))))))
-
-(define
-  mod/decide
-  (fn
-    (id)
-    (let
-      ((r (mod/get-report id)))
-      (if
-        (nil? r)
-        nil
-        (let
-          ((d (mod/decide-report r mod/*reports* mod/*rules*)))
-          (begin (mod/log-decision! d (mod/report-evidence r)) d))))))
-
-;; ── lifecycle façade over the case registry ──
-
-(define
-  mod/case-of
-  (fn
-    (id)
-    (reduce
-      (fn (acc rec) (if (= (get rec :id) id) (get rec :case) acc))
-      nil
-      mod/*cases*)))
-
-(define
-  mod/case-store!
-  (fn
-    (id c)
-    (set!
-      mod/*cases*
-      (map
-        (fn (rec) (if (= (get rec :id) id) {:id id :case c} rec))
-        mod/*cases*))))
-
-;; apply a lifecycle op to the stored case, persist it, and (when a decision was
-;; committed cleanly) append it to the audit log; returns the updated case
-(define
-  mod/case-apply!
-  (fn
-    (id op log?)
-    (let
-      ((c (mod/case-of id)))
-      (if
-        (nil? c)
-        nil
-        (let
-          ((c2 (op c)))
-          (begin
-            (mod/case-store! id c2)
-            (when
-              log?
-              (when
-                (nil? (mod/case-error c2))
-                (let
-                  ((d (mod/case-decision c2)))
-                  (if
-                    (nil? d)
-                    nil
-                    (mod/log-decision!
-                      d
-                      (mod/report-evidence (mod/case-report c2)))))))
-            c2))))))
-
-(define
-  mod/triage
-  (fn
-    (id)
-    (mod/case-apply!
-      id
-      (fn (c) (mod/case-triage c mod/*reports* mod/*rules*))
-      false)))
-
-(define
-  mod/resolve
-  (fn (id) (mod/case-apply! id (fn (c) (mod/case-resolve c)) true)))
-
-(define
-  mod/review
-  (fn
-    (id kind val)
-    (mod/case-apply!
-      id
-      (fn (c) (mod/case-review c kind val mod/*reports* mod/*rules*))
-      true)))
-
-(define
-  mod/appeal
-  (fn
-    (id kind val)
-    (mod/case-apply!
-      id
-      (fn (c) (mod/case-appeal c kind val mod/*reports* mod/*rules*))
-      true)))
-
-(define
-  mod/finalize
-  (fn (id) (mod/case-apply! id (fn (c) (mod/case-finalize c)) false)))

lib/mod/audit.sx

@@ -1,54 +0,0 @@
-;; lib/mod/audit.sx — append-only decision log.
-;;
-;; Every decision the api commits is recorded as an immutable audit entry holding
-;; the decision (action + matching rule), the proof tree (the derivation that
-;; justified it), and a snapshot of the evidence in force at decision time. The
-;; log is append-only: entries are never mutated or removed, only appended, each
-;; with a monotonic sequence number. Retrieval is by report id (full history) or
-;; by sequence.
-
-(define mod/*audit-log* (list))
-(define mod/*audit-seq* 0)
-
-(define
-  mod/audit-reset!
-  (fn
-    ()
-    (begin (set! mod/*audit-log* (list)) (set! mod/*audit-seq* 0))))
-
-(define mod/mk-audit-entry (fn (seq decision evidence-snapshot) {:action (get decision :action) :evidence evidence-snapshot :proof (get decision :proof) :rule (get decision :rule) :report-id (get decision :report-id) :seq seq}))
-
-(define
-  mod/log-decision!
-  (fn
-    (decision evidence-snapshot)
-    (begin
-      (set! mod/*audit-seq* (+ mod/*audit-seq* 1))
-      (let
-        ((entry (mod/mk-audit-entry mod/*audit-seq* decision evidence-snapshot)))
-        (begin (append! mod/*audit-log* entry) entry)))))
-
-;; entries for one report, in chronological (sequence) order
-(define
-  mod/audit
-  (fn
-    (id)
-    (reduce
-      (fn
-        (acc e)
-        (if (= (get e :report-id) id) (append acc (list e)) acc))
-      (list)
-      mod/*audit-log*)))
-
-(define mod/audit-all (fn () mod/*audit-log*))
-(define mod/audit-count (fn () (len mod/*audit-log*)))
-
-;; most recent decision logged for a report (nil if none)
-(define
-  mod/audit-latest
-  (fn
-    (id)
-    (reduce
-      (fn (acc e) (if (= (get e :report-id) id) e acc))
-      nil
-      mod/*audit-log*)))

lib/mod/batch.sx

@@ -1,55 +0,0 @@
-;; lib/mod/batch.sx — batch triage + corpus analytics.
-;;
-;; Operational layer: decide a whole queue of reports at once, summarize the
-;; outcomes by action, and measure which rules actually fire across a corpus.
-;; mod/never-fired is the empirical complement to lint's static unreachable check
-;; (Ext 5): lint finds rules that CAN'T fire by structure; never-fired finds rules
-;; that DIDN'T fire on real data.
-
-(define
-  mod/decide-batch
-  (fn
-    (reports rules)
-    (map (fn (r) (mod/decide-report r reports rules)) reports)))
-
-(define
-  mod/count-action
-  (fn
-    (decisions action)
-    (reduce
-      (fn (acc d) (if (= (get d :action) action) (+ acc 1) acc))
-      0
-      decisions)))
-
-(define mod/action-histogram (fn (decisions) {:keep (mod/count-action decisions "keep") :remove (mod/count-action decisions "remove") :escalate (mod/count-action decisions "escalate") :hide (mod/count-action decisions "hide") :ban (mod/count-action decisions "ban")}))
-
-(define
-  mod/rule-fire-count
-  (fn
-    (decisions rule-name)
-    (reduce
-      (fn (acc d) (if (= (get d :rule) rule-name) (+ acc 1) acc))
-      0
-      decisions)))
-
-(define
-  mod/rule-coverage
-  (fn
-    (reports rules)
-    (let
-      ((decisions (mod/decide-batch reports rules)))
-      (map (fn (rule) {:rule (mod/rule-name rule) :fired (mod/rule-fire-count decisions (mod/rule-name rule))}) rules))))
-
-(define
-  mod/never-fired
-  (fn
-    (reports rules)
-    (reduce
-      (fn
-        (acc c)
-        (if
-          (= (get c :fired) 0)
-          (append acc (list (get c :rule)))
-          acc))
-      (list)
-      (mod/rule-coverage reports rules))))

lib/mod/conformance.conf

@@ -1,60 +0,0 @@
-# Mod conformance config — sourced by lib/guest/conformance.sh.
-
-LANG_NAME=mod
-MODE=dict
-
-PRELOADS=(
-  lib/guest/pratt.sx
-  lib/prolog/tokenizer.sx
-  lib/prolog/parser.sx
-  lib/prolog/runtime.sx
-  lib/prolog/query.sx
-  lib/prolog/compiler.sx
-  lib/mod/schema.sx
-  lib/mod/policy.sx
-  lib/mod/defrule.sx
-  lib/mod/engine.sx
-  lib/mod/explain.sx
-  lib/mod/severity.sx
-  lib/mod/offenders.sx
-  lib/mod/quorum.sx
-  lib/mod/trace.sx
-  lib/mod/whatif.sx
-  lib/mod/batch.sx
-  lib/mod/temporal.sx
-  lib/mod/sla.sx
-  lib/mod/wire.sx
-  lib/mod/activity.sx
-  lib/mod/policies.sx
-  lib/mod/pipeline.sx
-  lib/mod/lifecycle.sx
-  lib/mod/audit.sx
-  lib/mod/api.sx
-  lib/mod/fed.sx
-  lib/mod/link.sx
-  lib/mod/lint.sx
-)
-
-SUITES=(
-  "decide:lib/mod/tests/decide.sx:(mod-decide-tests-run!)"
-  "audit:lib/mod/tests/audit.sx:(mod-audit-tests-run!)"
-  "escalation:lib/mod/tests/escalation.sx:(mod-escalation-tests-run!)"
-  "fed:lib/mod/tests/fed.sx:(mod-fed-tests-run!)"
-  "extensions:lib/mod/tests/extensions.sx:(mod-extensions-tests-run!)"
-  "link:lib/mod/tests/link.sx:(mod-link-tests-run!)"
-  "lint:lib/mod/tests/lint.sx:(mod-lint-tests-run!)"
-  "severity:lib/mod/tests/severity.sx:(mod-severity-tests-run!)"
-  "offenders:lib/mod/tests/offenders.sx:(mod-offenders-tests-run!)"
-  "quorum:lib/mod/tests/quorum.sx:(mod-quorum-tests-run!)"
-  "trace:lib/mod/tests/trace.sx:(mod-trace-tests-run!)"
-  "whatif:lib/mod/tests/whatif.sx:(mod-whatif-tests-run!)"
-  "batch:lib/mod/tests/batch.sx:(mod-batch-tests-run!)"
-  "temporal:lib/mod/tests/temporal.sx:(mod-temporal-tests-run!)"
-  "sla:lib/mod/tests/sla.sx:(mod-sla-tests-run!)"
-  "wire:lib/mod/tests/wire.sx:(mod-wire-tests-run!)"
-  "disjunction:lib/mod/tests/disjunction.sx:(mod-disjunction-tests-run!)"
-  "activity:lib/mod/tests/activity.sx:(mod-activity-tests-run!)"
-  "policies:lib/mod/tests/policies.sx:(mod-policies-tests-run!)"
-  "defrule:lib/mod/tests/defrule.sx:(mod-defrule-tests-run!)"
-  "pipeline:lib/mod/tests/pipeline.sx:(mod-pipeline-tests-run!)"
-)

lib/mod/conformance.sh

@@ -1,3 +0,0 @@
-#!/usr/bin/env bash
-# Thin wrapper — see lib/guest/conformance.sh and lib/mod/conformance.conf.
-exec bash "$(dirname "$0")/../guest/conformance.sh" "$(dirname "$0")/conformance.conf" "$@"

lib/mod/defrule.sx

@@ -1,16 +0,0 @@
-;; lib/mod/defrule.sx — ergonomic rule / ruleset construction.
-;;
-;; The roadmap sketched a (defrule action :when conditions) surface. Conditions
-;; already evaluate to plain data, so this needs no macro — variadic functions
-;; suffice: mod/defrule collects its trailing condition forms via &rest (dropping
-;; the explicit outer (list ...)), and mod/ruleset assembles rules the same way.
-;;
-;;   (mod/ruleset
-;;     (mod/defrule "spam-hide" :hide (list :classification "spam"))
-;;     (mod/defrule "default-keep" :keep))
-
-(define
-  mod/defrule
-  (fn (name action &rest conds) (mod/mk-rule name action conds)))
-
-(define mod/ruleset (fn (&rest rules) rules))

lib/mod/engine.sx

@@ -1,64 +0,0 @@
-;; lib/mod/engine.sx — decide a report by querying the policy program.
-;;
-;; build-program assembles the report's facts plus the compiled policy clauses;
-;; decide-report runs the Prolog query and returns a decision. A decision is a
-;; proof, not a bare keyword: it carries the matching rule, the conditions it
-;; required, the evidence that satisfied them, and a derivation — the proof tree.
-;;
-;; The proof tree is built constructively: for the matching rule, each body goal
-;; is re-queried against the same DB with the report id bound, recording the goal
-;; text, whether it was solved, and the bindings that satisfied it. That is a
-;; genuine derivation drawn from the Prolog database, ready for the audit trail.
-
-(define
-  mod/find-rule
-  (fn
-    (rules name)
-    (reduce
-      (fn
-        (acc r)
-        (if (nil? acc) (if (= (mod/rule-name r) name) r acc) acc))
-      nil
-      rules)))
-
-(define
-  mod/build-program
-  (fn
-    (r count rules)
-    (str (mod/report-facts r count) "\n" (mod/rules->program rules))))
-
-(define
-  mod/proof-goals
-  (fn
-    (db id conds)
-    (if
-      (empty? conds)
-      (list {:solved true :goal "true" :bindings {}})
-      (map
-        (fn
-          (c)
-          (let
-            ((g (mod/cond->goal c id)))
-            (let ((sols (pl-query-all db g))) {:solved (if (empty? sols) false true) :goal g :bindings (if (empty? sols) {} (first sols))})))
-        conds))))
-
-(define
-  mod/decide-report
-  (fn
-    (r reports rules)
-    (let
-      ((count (mod/report-count (mod/report-about r) reports))
-        (kinds (mod/classify-keywords r))
-        (id (mod/report-id r)))
-      (let
-        ((program (mod/build-program r count rules)))
-        (let
-          ((db (pl-load program)))
-          (let
-            ((sol (pl-query-one db (str "policy_action(" id ", Action, Rule)"))))
-            (if
-              (nil? sol)
-              {:action "keep" :proof {:goals (list) :evidence kinds :conditions (list) :rule "none" :count count} :report-id id :rule "none"}
-              (let
-                ((rname (dict-get sol "Rule")))
-                (let ((rule (mod/find-rule rules rname))) {:action (mod/rule-action rule) :proof {:goals (mod/proof-goals db id (mod/rule-when rule)) :evidence kinds :conditions (mod/rule-when rule) :rule rname :count count} :report-id id :rule rname})))))))))

lib/mod/explain.sx

@@ -1,55 +0,0 @@
-;; lib/mod/explain.sx — human-readable proof explanation.
-;;
-;; Turns a decision (from mod/decide-report, or any audit entry) into a readable
-;; multi-line "why": the action, the rule that fired, the evidence in play, and
-;; the derivation goal-by-goal with [proved]/[unproved] marks and the unification
-;; bindings that satisfied each goal. Pure SX over the Phase-2 proof tree.
-
-(define
-  mod/explain-binds
-  (fn
-    (binds)
-    (mod/join-with
-      ", "
-      (map (fn (k) (str k "=" (dict-get binds k))) (keys binds)))))
-
-(define
-  mod/explain-goal
-  (fn
-    (g)
-    (let
-      ((mark (if (get g :solved) "  [proved] " "  [unproved] "))
-        (binds (get g :bindings)))
-      (if
-        (empty? (keys binds))
-        (str mark (get g :goal))
-        (str mark (get g :goal) "  {" (mod/explain-binds binds) "}")))))
-
-(define
-  mod/explain-evidence
-  (fn
-    (evidence)
-    (if
-      (empty? evidence)
-      "Evidence: (none)"
-      (str "Evidence: " (mod/join-with ", " evidence)))))
-
-(define
-  mod/explain
-  (fn
-    (decision)
-    (let
-      ((id (get decision :report-id))
-        (action (get decision :action))
-        (rule (get decision :rule))
-        (proof (get decision :proof)))
-      (let
-        ((goals (get proof :goals)) (evidence (get proof :evidence)))
-        (mod/join-with
-          "\n"
-          (append
-            (list
-              (str "Report " id ": " action " (rule: " rule ")")
-              (mod/explain-evidence evidence)
-              "Because:")
-            (map mod/explain-goal goals)))))))

lib/mod/fed.sx

@@ -1,145 +0,0 @@
-;; lib/mod/fed.sx — federation: cross-instance reports, decision sharing, trust,
-;; revocation. fed-sx itself is mocked here (an in-memory outbox); the real wire
-;; transport would replace mod/fed-send!.
-;;
-;; Trust is advisory by default (the hard rule): a peer's decision only binds
-;; locally when (mod/trusted? peer :mod) holds. An untrusted peer's decision is
-;; recorded as a suggestion in the advisory log and is NOT applied. Local
-;; decisions propagate outward via the outbox. Revocation undoes a locally
-;; applied action when its proof is invalidated, notifying the origin peer.
-
-(define mod/*fed-trust* (list))     ;; {:peer :scope}
-(define mod/*fed-outbox* (list))    ;; {:to :type :payload}
-(define mod/*fed-advisory* (list))  ;; {:peer :decision} — received, not applied
-(define mod/*fed-applied* (list))   ;; {:report-id :action :origin :revoked}
-(define mod/*fed-origins* (list))   ;; {:id :origin}
-
-(define
-  mod/fed-reset!
-  (fn
-    ()
-    (begin
-      (set! mod/*fed-trust* (list))
-      (set! mod/*fed-outbox* (list))
-      (set! mod/*fed-advisory* (list))
-      (set! mod/*fed-applied* (list))
-      (set! mod/*fed-origins* (list)))))
-
-;; ── trust model ──
-
-(define
-  mod/trust-match?
-  (fn
-    (t peer scope)
-    (if (= (get t :peer) peer) (= (get t :scope) scope) false)))
-
-(define
-  mod/grant-trust
-  (fn (peer scope) (begin (append! mod/*fed-trust* {:scope scope :peer peer}) true)))
-
-(define
-  mod/revoke-trust
-  (fn
-    (peer scope)
-    (set!
-      mod/*fed-trust*
-      (reduce
-        (fn
-          (acc t)
-          (if (mod/trust-match? t peer scope) acc (append acc (list t))))
-        (list)
-        mod/*fed-trust*))))
-
-(define
-  mod/trusted?
-  (fn
-    (peer scope)
-    (mod/any? (fn (t) (mod/trust-match? t peer scope)) mod/*fed-trust*)))
-
-;; ── cross-instance reports ──
-
-(define
-  mod/fed-receive-report
-  (fn
-    (peer by about reason)
-    (let
-      ((r (mod/report by about reason)))
-      (begin (append! mod/*fed-origins* {:id (mod/report-id r) :origin peer}) r))))
-
-(define
-  mod/report-origin
-  (fn
-    (id)
-    (reduce
-      (fn (acc o) (if (= (get o :id) id) (get o :origin) acc))
-      "local"
-      mod/*fed-origins*)))
-
-;; ── decision sharing (mock fed-sx send) ──
-
-(define
-  mod/fed-send!
-  (fn (to type payload) (begin (append! mod/*fed-outbox* {:type type :to to :payload payload}) true)))
-
-(define mod/fed-outbox (fn () mod/*fed-outbox*))
-
-(define
-  mod/fed-share-decision
-  (fn
-    (decision peers)
-    (reduce
-      (fn
-        (acc p)
-        (begin (mod/fed-send! p "decision" decision) (append acc (list p))))
-      (list)
-      peers)))
-
-;; ── receiving a peer's decision (advisory unless trusted) ──
-
-(define
-  mod/fed-applied-action
-  (fn
-    (report-id)
-    (reduce
-      (fn (acc a) (if (= (get a :report-id) report-id) a acc))
-      nil
-      mod/*fed-applied*)))
-
-(define
-  mod/fed-receive-decision
-  (fn
-    (peer decision)
-    (if
-      (mod/trusted? peer :mod)
-      (begin (append! mod/*fed-applied* {:revoked false :action (get decision :action) :report-id (get decision :report-id) :origin peer}) {:advisory false :peer peer :applied true :decision decision})
-      (begin (append! mod/*fed-advisory* {:peer peer :decision decision}) {:advisory true :peer peer :applied false :decision decision}))))
-
-;; ── revocation ──
-
-(define
-  mod/fed-revoke!
-  (fn
-    (report-id reason)
-    (begin
-      (set!
-        mod/*fed-applied*
-        (map
-          (fn (a) (if (= (get a :report-id) report-id) {:revoked true :action (get a :action) :report-id (get a :report-id) :origin (get a :origin)} a))
-          mod/*fed-applied*))
-      (mod/fed-send! (mod/report-origin report-id) "revocation" {:report-id report-id :reason reason})
-      report-id)))
-
-;; re-run the engine; if the action no longer holds, the prior decision's proof
-;; is invalidated — revoke the applied moderation.
-(define
-  mod/fed-revoke-if-invalidated
-  (fn
-    (report decision reports rules)
-    (let
-      ((d2 (mod/decide-report report reports rules)))
-      (if
-        (= (get d2 :action) (get decision :action))
-        {:revoked false :decision d2}
-        (begin
-          (mod/fed-revoke! (get decision :report-id) "proof invalidated")
-          {:revoked true :decision d2})))))

lib/mod/lifecycle.sx

@@ -1,160 +0,0 @@
-;; lib/mod/lifecycle.sx — report lifecycle state machine (pure SX over the engine).
-;;
-;; Lifecycle state is deliberately separate from policy: the Prolog rules answer
-;; "what action?", this module answers "where in the process is this report?".
-;;
-;;   :open ──triage──▶ :triaged ──resolve/review──▶ :decided ──appeal──▶ :appealed
-;;                                                      │                    │
-;;                                                      └────finalize───▶ :final ◀┘
-;;
-;; A case is an immutable value {:report :state :decision :tier :error :history}.
-;; Every transition returns a NEW case; illegal transitions return the case
-;; unchanged with :error set. Tiers: triage runs the engine (auto-tier); a
-;; terminal action (hide/remove/keep) resolves immediately, an :escalate action
-;; flags the case for human review (human-tier) before it can be resolved.
-
-(define mod/case* (fn (report state decision tier err history) {:history history :state state :report report :error err :tier tier :decision decision}))
-
-(define
-  mod/mk-case
-  (fn (report) (mod/case* report "open" nil nil nil (list))))
-
-(define mod/case-report (fn (c) (get c :report)))
-(define mod/case-state (fn (c) (get c :state)))
-(define mod/case-decision (fn (c) (get c :decision)))
-(define mod/case-tier (fn (c) (get c :tier)))
-(define mod/case-error (fn (c) (get c :error)))
-(define mod/case-history (fn (c) (get c :history)))
-
-;; ── transition table ──
-
-(define mod/lc-transitions {:final (list) :appealed (list "final") :decided (list "appealed" "final") :open (list "triaged") :triaged (list "decided")})
-
-(define mod/member? (fn (x lst) (mod/any? (fn (y) (= y x)) lst)))
-
-(define
-  mod/lc-can-transition?
-  (fn
-    (from to)
-    (let
-      ((outs (get mod/lc-transitions from)))
-      (if (nil? outs) false (mod/member? to outs)))))
-
-;; ── core transition: validate, record history, or flag :error ──
-
-(define
-  mod/case-goto
-  (fn
-    (c to note report decision tier)
-    (let
-      ((from (mod/case-state c)))
-      (if
-        (mod/lc-can-transition? from to)
-        (mod/case*
-          report
-          to
-          decision
-          tier
-          nil
-          (append (mod/case-history c) (list {:note note :to to :from from})))
-        (mod/case*
-          (mod/case-report c)
-          from
-          (mod/case-decision c)
-          (mod/case-tier c)
-          (str "illegal transition: " from " -> " to)
-          (mod/case-history c))))))
-
-(define
-  mod/case-error-set
-  (fn
-    (c msg)
-    (mod/case*
-      (mod/case-report c)
-      (mod/case-state c)
-      (mod/case-decision c)
-      (mod/case-tier c)
-      msg
-      (mod/case-history c))))
-
-;; ── lifecycle operations ──
-
-;; :open → :triaged — run the auto-tier first pass.
-(define
-  mod/case-triage
-  (fn
-    (c reports rules)
-    (let
-      ((d (mod/decide-report (mod/case-report c) reports rules)))
-      (let
-        ((tier (if (= (get d :action) "escalate") "human" "auto")))
-        (mod/case-goto
-          c
-          "triaged"
-          "auto-tier first pass"
-          (mod/case-report c)
-          d
-          tier)))))
-
-;; :triaged → :decided — auto-tier resolves; human-tier is blocked until review.
-(define
-  mod/case-resolve
-  (fn
-    (c)
-    (if
-      (= (mod/case-tier c) "human")
-      (mod/case-error-set c "awaiting human review (escalated)")
-      (mod/case-goto
-        c
-        "decided"
-        "auto-tier resolved"
-        (mod/case-report c)
-        (mod/case-decision c)
-        (mod/case-tier c)))))
-
-;; :triaged → :decided — human review: attach evidence, re-decide, resolve.
-(define
-  mod/case-review
-  (fn
-    (c kind val reports rules)
-    (let
-      ((nr (mod/attach-evidence (mod/case-report c) (mod/mk-evidence kind val))))
-      (let
-        ((d (mod/decide-report nr reports rules)))
-        (mod/case-goto c "decided" (str "human review: " kind) nr d "human")))))
-
-;; :decided → :appealed — appeal: attach evidence, re-decide (may override).
-(define
-  mod/case-appeal
-  (fn
-    (c kind val reports rules)
-    (let
-      ((nr (mod/attach-evidence (mod/case-report c) (mod/mk-evidence kind val))))
-      (let
-        ((d (mod/decide-report nr reports rules)))
-        (mod/case-goto
-          c
-          "appealed"
-          (str "appeal: " kind)
-          nr
-          d
-          (mod/case-tier c))))))
-
-;; :decided | :appealed → :final
-(define
-  mod/case-finalize
-  (fn
-    (c)
-    (mod/case-goto
-      c
-      "final"
-      "finalized"
-      (mod/case-report c)
-      (mod/case-decision c)
-      (mod/case-tier c))))
-
-(define
-  mod/case-action
-  (fn
-    (c)
-    (let ((d (mod/case-decision c))) (if (nil? d) nil (get d :action)))))

lib/mod/link.sx

@@ -1,92 +0,0 @@
-;; lib/mod/link.sx — report linking + deduplication.
-;;
-;; Reports about the same subject form a cluster; identical reports (same
-;; reporter + subject + reason) are duplicates. Linking is Prolog-backed: all
-;; report facts are loaded and related ids are found by unification — the same
-;; relational substrate the policy engine uses, here for retrieval rather than
-;; decision. Dedup is pure SX over a normalized link key.
-
-(define
-  mod/link-key
-  (fn
-    (r)
-    (str
-      (mod/report-by r)
-      "|"
-      (mod/report-about r)
-      "|"
-      (downcase (mod/report-reason r)))))
-
-(define
-  mod/dedup-reports
-  (fn
-    (reports)
-    (reduce
-      (fn
-        (acc r)
-        (if
-          (mod/any? (fn (x) (= (mod/link-key x) (mod/link-key r))) acc)
-          acc
-          (append acc (list r))))
-      (list)
-      reports)))
-
-(define
-  mod/duplicate-count
-  (fn (reports) (- (len reports) (len (mod/dedup-reports reports)))))
-
-;; ── Prolog-backed relational retrieval ──
-
-(define
-  mod/report-rel-facts
-  (fn
-    (reports)
-    (mod/join-with
-      "\n"
-      (map
-        (fn
-          (r)
-          (str
-            "report("
-            (mod/report-id r)
-            ", "
-            (mod/pl-quote (mod/report-by r))
-            ", "
-            (mod/pl-quote (mod/report-about r))
-            ")."))
-        reports))))
-
-(define
-  mod/related-ids
-  (fn
-    (subject reports)
-    (let
-      ((db (pl-load (mod/report-rel-facts reports))))
-      (map
-        (fn (sol) (dict-get sol "Id"))
-        (pl-query-all db (str "report(Id, _, " (mod/pl-quote subject) ")"))))))
-
-(define
-  mod/reporters-of
-  (fn
-    (subject reports)
-    (let
-      ((db (pl-load (mod/report-rel-facts reports))))
-      (map
-        (fn (sol) (dict-get sol "By"))
-        (pl-query-all db (str "report(_, By, " (mod/pl-quote subject) ")"))))))
-
-(define
-  mod/distinct
-  (fn
-    (items)
-    (reduce
-      (fn
-        (acc x)
-        (if (mod/any? (fn (y) (= y x)) acc) acc (append acc (list x))))
-      (list)
-      items)))
-
-(define
-  mod/distinct-reporters-of
-  (fn (subject reports) (mod/distinct (mod/reporters-of subject reports))))

lib/mod/lint.sx

@@ -1,69 +0,0 @@
-;; lib/mod/lint.sx — static analysis of a policy rule set.
-;;
-;; Because precedence is "first matching clause wins" (pl-query-one), the rule
-;; order has correctness consequences a moderator can get wrong: a rule placed
-;; after an unconditional (empty :when) rule can never fire, and a rule set with
-;; no unconditional rule may leave some reports undecided. lint-rules surfaces
-;; these without running the engine.
-
-(define mod/rule-unconditional? (fn (r) (empty? (mod/rule-when r))))
-
-;; names of rules that follow the first unconditional rule — structurally dead,
-;; since the unconditional rule always matches first
-(define
-  mod/unreachable-rules
-  (fn
-    (rules)
-    (get
-      (reduce
-        (fn
-          (acc r)
-          (if
-            (get acc :hit)
-            {:dead (append (get acc :dead) (list (mod/rule-name r))) :hit true}
-            (if (mod/rule-unconditional? r) {:dead (get acc :dead) :hit true} acc)))
-        {:dead (list) :hit false}
-        rules)
-      :dead)))
-
-(define
-  mod/has-catchall?
-  (fn (rules) (mod/any? mod/rule-unconditional? rules)))
-
-(define
-  mod/count-eq
-  (fn
-    (x lst)
-    (reduce (fn (a y) (if (= y x) (+ a 1) a)) 0 lst)))
-
-(define
-  mod/duplicate-rule-names
-  (fn
-    (rules)
-    (let
-      ((names (map mod/rule-name rules)))
-      (mod/distinct
-        (reduce
-          (fn
-            (acc n)
-            (if
-              (< 1 (mod/count-eq n names))
-              (append acc (list n))
-              acc))
-          (list)
-          names)))))
-
-(define mod/lint-rules (fn (rules) {:duplicate-names (mod/duplicate-rule-names rules) :has-catchall (mod/has-catchall? rules) :unreachable (mod/unreachable-rules rules)}))
-
-;; a rule set is well-formed when nothing is dead, it has a catch-all, and rule
-;; names are unique
-(define
-  mod/rules-ok?
-  (fn
-    (rules)
-    (let
-      ((l (mod/lint-rules rules)))
-      (if
-        (empty? (get l :unreachable))
-        (if (get l :has-catchall) (empty? (get l :duplicate-names)) false)
-        false))))

lib/mod/offenders.sx

@@ -1,59 +0,0 @@
-;; lib/mod/offenders.sx — repeat-offender escalation (audit log as evidence).
-;;
-;; The append-only audit trail is itself a source of evidence: a subject already
-;; sanctioned several times is a repeat offender. mod/decide-escalating decides a
-;; report normally, then — if the action is a sanction and the subject has at
-;; least k PRIOR sanctions in the audit log — upgrades it to :ban. This is the one
-;; place a decision depends on history beyond the single report, and it reads that
-;; history from the audit log rather than re-deriving it.
-
-(define
-  mod/sanction?
-  (fn
-    (action)
-    (mod/any? (fn (a) (= a action)) (list "hide" "remove" "ban"))))
-
-;; count of prior sanctioning decisions in the audit log about a subject
-(define
-  mod/subject-sanctions
-  (fn
-    (subject)
-    (reduce
-      (fn
-        (acc e)
-        (let
-          ((r (mod/get-report (get e :report-id))))
-          (if
-            (nil? r)
-            acc
-            (if
-              (if
-                (= (mod/report-about r) subject)
-                (mod/sanction? (get e :action))
-                false)
-              (+ acc 1)
-              acc))))
-      0
-      (mod/audit-all))))
-
-(define
-  mod/repeat-offender?
-  (fn (subject k) (<= k (mod/subject-sanctions subject))))
-
-(define
-  mod/decide-escalating
-  (fn
-    (id k)
-    (let
-      ((r (mod/get-report id)))
-      (if
-        (nil? r)
-        nil
-        (let
-          ((priors (mod/subject-sanctions (mod/report-about r))))
-          (let
-            ((d (mod/decide id)))
-            (if
-              (if (mod/sanction? (get d :action)) (<= k priors) false)
-              {:action "ban" :proof {:goals (get (get d :proof) :goals) :prior-sanctions priors :evidence (get (get d :proof) :evidence) :conditions (list) :rule "repeat-offender-ban" :count (get (get d :proof) :count)} :report-id id :rule "repeat-offender-ban" :strategy "escalating"}
-              d)))))))

lib/mod/pipeline.sx

@@ -1,18 +0,0 @@
-;; lib/mod/pipeline.sx — end-to-end triage orchestration.
-;;
-;; A single entry point that runs a report through the subsystem and returns the
-;; full artifact bundle: the decision (under the report's domain policy), a
-;; human-readable explanation, an ActivityPub-shaped event for the bus, and the
-;; wire line for federated peers. Composes policies (Ext 17), explain (Ext 3),
-;; activity (Ext 16) and wire (Ext 14) — the modules are independent, this is just
-;; the convenience that wires them together for the common "process a report" path.
-
-(define
-  mod/triage-pipeline
-  (fn
-    (domain r reports actor)
-    (let ((d (mod/decide-in domain r reports))) {:activity (mod/decision->activity d actor) :action (get d :action) :wire (mod/decision->wire d) :rule (get d :rule) :decision d :explanation (mod/explain d)})))
-
-(define mod/pipeline-action (fn (p) (get p :action)))
-(define mod/pipeline-activity (fn (p) (get p :activity)))
-(define mod/pipeline-wire (fn (p) (get p :wire)))

lib/mod/policies.sx

@@ -1,40 +0,0 @@
-;; lib/mod/policies.sx — per-domain policy registry.
-;;
-;; rose-ash spans domains (blog, market, events, federation, …) that want
-;; different moderation — a marketplace listing and a blog comment are not held to
-;; the same bar. This registry maps a domain to a rule set; mod/decide-in resolves
-;; the right policy and decides. Unregistered domains fall back to the default
-;; rules, so adding a domain never leaves it unmoderated.
-
-(define mod/*policies* (list))
-
-(define mod/policies-reset! (fn () (set! mod/*policies* (list))))
-
-(define
-  mod/register-policy!
-  (fn (domain rules) (begin (append! mod/*policies* {:domain domain :rules rules}) true)))
-
-(define
-  mod/policy-registered?
-  (fn
-    (domain)
-    (mod/any? (fn (p) (= (get p :domain) domain)) mod/*policies*)))
-
-(define
-  mod/policy-for
-  (fn
-    (domain)
-    (reduce
-      (fn (acc p) (if (= (get p :domain) domain) (get p :rules) acc))
-      mod/default-rules
-      mod/*policies*)))
-
-(define
-  mod/decide-in
-  (fn
-    (domain r reports)
-    (mod/decide-report r reports (mod/policy-for domain))))
-
-(define
-  mod/registered-domains
-  (fn () (map (fn (p) (get p :domain)) mod/*policies*)))

lib/mod/policy.sx

@@ -1,137 +0,0 @@
-;; lib/mod/policy.sx — moderation rules → Prolog clauses.
-;;
-;; A rule is {:name :action :when}. :when is a list of condition forms; each
-;; compiles to a Prolog goal. The conditions in a :when list are ANDed (joined by
-;; ", "); :not negates and :any (a list of sub-conditions) disjoins — so the
-;; condition language is a small boolean algebra over the leaf predicates.
-;; Rule order is precedence: the engine queries with pl-query-one, so the first
-;; clause that proves wins. The final default rule has an empty body (true) so
-;; every report yields at least :keep — "no rule matched" is a real result, not a
-;; query failure.
-;;
-;; cond->goal takes an id-term so the same condition can be compiled with the
-;; head variable "Id" (for clause bodies) or a concrete report id (for proof-tree
-;; goal-by-goal re-querying in the engine).
-;;
-;; Precedence (top wins): exoneration evidence (appeal override) > confirmed-abuse
-;; evidence (human review) > spam/abuse classification > repeated-report count >
-;; default keep.
-
-(define mod/mk-rule (fn (name action conds) {:when conds :name name :action action}))
-
-(define mod/rule-name (fn (r) (get r :name)))
-(define mod/rule-action (fn (r) (get r :action)))
-(define mod/rule-when (fn (r) (get r :when)))
-
-(define
-  mod/default-rules
-  (list
-    (mod/mk-rule
-      "exonerated-keep"
-      :keep (list (list :evidence "exonerated")))
-    (mod/mk-rule
-      "reviewer-remove"
-      :remove (list (list :evidence "confirmed-abuse")))
-    (mod/mk-rule "spam-hide" :hide (list (list :classification "spam")))
-    (mod/mk-rule
-      "abuse-remove"
-      :remove (list (list :classification "abuse")))
-    (mod/mk-rule
-      "repeated-escalate"
-      :escalate (list (list :count-at-least 3)))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-;; ── condition → Prolog goal ──
-;;
-;; (:classification "spam")  → classification(Id, spam)
-;; (:evidence "kind")        → evidence(Id, 'kind', _)
-;; (:attr "verified")        → attr(Id, verified)
-;; (:not <cond>)             → not(<cond>)                        (negation)
-;; (:any (list c1 c2 ...))   → (g1 ; g2 ; ...)                    (disjunction)
-;; (:count-at-least 3)       → report(Id, B, S), report_count(S, N), N >= 3
-;; (:score-at-least 5)       → aggregate_all(sum(W), signal(Id, _, W), T), T >= 5
-;; (:reporters-at-least 2)   → report(Id, _, Sr), setof(Br, report(_, Br, Sr), Bsr),
-;;                             length(Bsr, Nr), Nr >= 2   (quorum engine)
-;; (:burst-at-least 3)       → report(Id, _, Sb), burst_count(Sb, Nb), Nb >= 3
-;;                             (temporal engine)
-
-(define
-  mod/cond->goal
-  (fn
-    (c idterm)
-    (let
-      ((tag (first c)))
-      (cond
-        ((= tag :classification)
-          (str "classification(" idterm ", " (nth c 1) ")"))
-        ((= tag :evidence)
-          (str
-            "evidence("
-            idterm
-            ", "
-            (mod/pl-quote (nth c 1))
-            ", _)"))
-        ((= tag :attr) (str "attr(" idterm ", " (nth c 1) ")"))
-        ((= tag :not)
-          (str "not(" (mod/cond->goal (nth c 1) idterm) ")"))
-        ((= tag :any)
-          (str
-            "("
-            (mod/join-with
-              " ; "
-              (map
-                (fn (sub) (mod/cond->goal sub idterm))
-                (nth c 1)))
-            ")"))
-        ((= tag :count-at-least)
-          (str
-            "report("
-            idterm
-            ", B, S), report_count(S, N), N >= "
-            (nth c 1)))
-        ((= tag :score-at-least)
-          (str
-            "aggregate_all(sum(W), signal("
-            idterm
-            ", _, W), T), T >= "
-            (nth c 1)))
-        ((= tag :reporters-at-least)
-          (str
-            "report("
-            idterm
-            ", _, Sr), setof(Br, report(_, Br, Sr), Bsr), "
-            "length(Bsr, Nr), Nr >= "
-            (nth c 1)))
-        ((= tag :burst-at-least)
-          (str
-            "report("
-            idterm
-            ", _, Sb), burst_count(Sb, Nb), Nb >= "
-            (nth c 1)))
-        (true "true")))))
-
-(define
-  mod/conds->body
-  (fn
-    (conds idterm)
-    (if
-      (empty? conds)
-      "true"
-      (mod/join-with ", " (map (fn (c) (mod/cond->goal c idterm)) conds)))))
-
-(define
-  mod/rule->clause
-  (fn
-    (r)
-    (str
-      "policy_action(Id, "
-      (mod/rule-action r)
-      ", '"
-      (mod/rule-name r)
-      "') :- "
-      (mod/conds->body (mod/rule-when r) "Id")
-      ".")))
-
-(define
-  mod/rules->program
-  (fn (rules) (mod/join-with "\n" (map mod/rule->clause rules))))

lib/mod/quorum.sx

@@ -1,40 +0,0 @@
-;; lib/mod/quorum.sx — quorum decisions over distinct reporters (anti-brigade).
-;;
-;; The base engine asserts only the decided report's report/3 fact, so it can't
-;; reason about WHO reported a subject. The quorum engine additionally asserts
-;; every report's report/3 fact (via link's rel-facts), letting a rule require N
-;; *distinct* reporters with `setof`/`length` — so one user filing many reports
-;; does not manufacture consensus. Same decision shape as the base engine, plus
-;; :strategy "quorum".
-
-(define
-  mod/build-quorum-program
-  (fn
-    (r count reports rules)
-    (str
-      (mod/report-rel-facts reports)
-      "\n"
-      (mod/report-facts r count)
-      "\n"
-      (mod/rules->program rules))))
-
-(define
-  mod/decide-quorum
-  (fn
-    (r reports rules)
-    (let
-      ((count (mod/report-count (mod/report-about r) reports))
-        (kinds (mod/classify-keywords r))
-        (id (mod/report-id r)))
-      (let
-        ((program (mod/build-quorum-program r count reports rules)))
-        (let
-          ((db (pl-load program)))
-          (let
-            ((sol (pl-query-one db (str "policy_action(" id ", Action, Rule)"))))
-            (if
-              (nil? sol)
-              {:action "keep" :proof {:goals (list) :evidence kinds :conditions (list) :rule "none" :count count} :report-id id :rule "none" :strategy "quorum"}
-              (let
-                ((rule (mod/find-rule rules (dict-get sol "Rule"))))
-                {:action (mod/rule-action rule) :proof {:goals (mod/proof-goals db id (mod/rule-when rule)) :evidence kinds :conditions (mod/rule-when rule) :rule (mod/rule-name rule) :count count} :report-id id :rule (mod/rule-name rule) :strategy "quorum"}))))))))

lib/mod/schema.sx

@@ -1,259 +0,0 @@
-;; lib/mod/schema.sx — report representation + Prolog fact generation.
-;;
-;; A report is a dict {:id :by :about :reason :evidence :attrs :signals :at}.
-;;   :evidence — accumulated {:kind :val} entries (human review, scanners)
-;;   :attrs    — attribute names ("verified") for negation-as-failure conditions
-;;   :signals  — weighted {:kind :weight} entries for aggregate scoring rules
-;;   :at       — integer timestamp/tick (deterministic; supplied, not clock-read)
-;; The engine derives keyword classifications from the reason text and projects
-;; the report, its classifications, evidence, attributes, and signals into Prolog
-;; facts that policy clauses match against.
-
-(define mod/mk-report (fn (id by about reason) {:attrs (list) :id id :signals (list) :by by :evidence (list) :about about :at 0 :reason reason}))
-
-(define mod/report-id (fn (r) (get r :id)))
-(define mod/report-by (fn (r) (get r :by)))
-(define mod/report-about (fn (r) (get r :about)))
-(define mod/report-reason (fn (r) (get r :reason)))
-
-(define
-  mod/report-evidence
-  (fn (r) (let ((e (get r :evidence))) (if (nil? e) (list) e))))
-
-(define
-  mod/report-attrs
-  (fn (r) (let ((a (get r :attrs))) (if (nil? a) (list) a))))
-
-(define
-  mod/report-signals
-  (fn (r) (let ((s (get r :signals))) (if (nil? s) (list) s))))
-
-(define
-  mod/report-at
-  (fn (r) (let ((t (get r :at))) (if (nil? t) 0 t))))
-
-(define mod/mk-evidence (fn (kind val) {:val val :kind kind}))
-(define mod/evidence-kind (fn (e) (get e :kind)))
-(define mod/evidence-val (fn (e) (get e :val)))
-
-(define mod/mk-signal (fn (kind weight) {:kind kind :weight weight}))
-(define mod/signal-kind (fn (s) (get s :kind)))
-(define mod/signal-weight (fn (s) (get s :weight)))
-
-(define mod/report* (fn (r evs attrs sigs at) {:attrs attrs :id (mod/report-id r) :signals sigs :by (mod/report-by r) :evidence evs :about (mod/report-about r) :at at :reason (mod/report-reason r)}))
-
-(define
-  mod/with-evidence
-  (fn
-    (r evs)
-    (mod/report*
-      r
-      evs
-      (mod/report-attrs r)
-      (mod/report-signals r)
-      (mod/report-at r))))
-
-(define
-  mod/with-attrs
-  (fn
-    (r attrs)
-    (mod/report*
-      r
-      (mod/report-evidence r)
-      attrs
-      (mod/report-signals r)
-      (mod/report-at r))))
-
-(define
-  mod/with-signals
-  (fn
-    (r sigs)
-    (mod/report*
-      r
-      (mod/report-evidence r)
-      (mod/report-attrs r)
-      sigs
-      (mod/report-at r))))
-
-(define
-  mod/with-at
-  (fn
-    (r at)
-    (mod/report*
-      r
-      (mod/report-evidence r)
-      (mod/report-attrs r)
-      (mod/report-signals r)
-      at)))
-
-(define
-  mod/attach-evidence
-  (fn
-    (r e)
-    (mod/with-evidence r (append (mod/report-evidence r) (list e)))))
-
-(define
-  mod/attach-attr
-  (fn (r a) (mod/with-attrs r (append (mod/report-attrs r) (list a)))))
-
-(define
-  mod/attach-signal
-  (fn (r s) (mod/with-signals r (append (mod/report-signals r) (list s)))))
-
-;; ── substring search (the prolog-loaded env lacks includes?; slice/len do work) ──
-
-(define
-  mod/contains-at?
-  (fn
-    (hay needle hl nl pos)
-    (if
-      (< hl (+ pos nl))
-      false
-      (if
-        (= (slice hay pos (+ pos nl)) needle)
-        true
-        (mod/contains-at? hay needle hl nl (+ pos 1))))))
-
-(define
-  mod/str-contains?
-  (fn
-    (hay needle)
-    (let
-      ((hl (len hay)) (nl (len needle)))
-      (if
-        (= nl 0)
-        true
-        (mod/contains-at? hay needle hl nl 0)))))
-
-;; ── evidence derivation (keyword classification) ──
-
-(define
-  mod/spam-keywords
-  (list "spam" "buy now" "click here" "free money" "viagra" "limited offer"))
-
-(define
-  mod/abuse-keywords
-  (list "abuse" "harassment" "threat" "slur" "hate speech"))
-
-(define
-  mod/any?
-  (fn (pred coll) (reduce (fn (acc x) (if acc acc (pred x))) false coll)))
-
-(define
-  mod/reason-matches?
-  (fn
-    (reason kws)
-    (let
-      ((low (downcase reason)))
-      (mod/any? (fn (k) (mod/str-contains? low k)) kws))))
-
-(define
-  mod/classify-keywords
-  (fn
-    (r)
-    (let
-      ((reason (mod/report-reason r)) (kinds (list)))
-      (begin
-        (when
-          (mod/reason-matches? reason mod/spam-keywords)
-          (append! kinds "spam"))
-        (when
-          (mod/reason-matches? reason mod/abuse-keywords)
-          (append! kinds "abuse"))
-        kinds))))
-
-(define
-  mod/report-count
-  (fn
-    (about reports)
-    (reduce
-      (fn
-        (acc r)
-        (if (= (mod/report-about r) about) (+ acc 1) acc))
-      0
-      reports)))
-
-;; ── Prolog fact projection ──
-
-(define
-  mod/join-with
-  (fn
-    (sep items)
-    (reduce (fn (acc x) (if (= acc "") x (str acc sep x))) "" items)))
-
-(define mod/pl-quote (fn (s) (str "'" s "'")))
-
-(define
-  mod/classification-facts
-  (fn
-    (id kinds)
-    (mod/join-with
-      "\n"
-      (map (fn (k) (str "classification(" id ", " k ").")) kinds))))
-
-(define
-  mod/evidence-facts
-  (fn
-    (id evs)
-    (mod/join-with
-      "\n"
-      (map
-        (fn
-          (e)
-          (str
-            "evidence("
-            id
-            ", "
-            (mod/pl-quote (mod/evidence-kind e))
-            ", "
-            (mod/pl-quote (str (mod/evidence-val e)))
-            ")."))
-        evs))))
-
-(define
-  mod/attr-facts
-  (fn
-    (id attrs)
-    (mod/join-with "\n" (map (fn (a) (str "attr(" id ", " a ").")) attrs))))
-
-(define
-  mod/signal-facts
-  (fn
-    (id sigs)
-    (mod/join-with
-      "\n"
-      (map
-        (fn
-          (s)
-          (str
-            "signal("
-            id
-            ", "
-            (mod/pl-quote (mod/signal-kind s))
-            ", "
-            (mod/signal-weight s)
-            ")."))
-        sigs))))
-
-(define
-  mod/report-facts
-  (fn
-    (r count)
-    (let
-      ((id (mod/report-id r))
-        (by (mod/pl-quote (mod/report-by r)))
-        (about (mod/pl-quote (mod/report-about r))))
-      (let
-        ((cls (mod/classification-facts id (mod/classify-keywords r)))
-          (evs (mod/evidence-facts id (mod/report-evidence r)))
-          (ats (mod/attr-facts id (mod/report-attrs r)))
-          (sgs (mod/signal-facts id (mod/report-signals r))))
-        (mod/join-with
-          "\n"
-          (list
-            (str "report(" id ", " by ", " about ").")
-            (str "report_count(" about ", " count ").")
-            cls
-            evs
-            ats
-            sgs))))))

lib/mod/scoreboard.json

@@ -1,30 +0,0 @@
-{
-  "lang": "mod",
-  "total_passed": 390,
-  "total_failed": 0,
-  "total": 390,
-  "suites": [
-    {"name":"decide","passed":31,"failed":0,"total":31},
-    {"name":"audit","passed":29,"failed":0,"total":29},
-    {"name":"escalation","passed":46,"failed":0,"total":46},
-    {"name":"fed","passed":26,"failed":0,"total":26},
-    {"name":"extensions","passed":32,"failed":0,"total":32},
-    {"name":"link","passed":12,"failed":0,"total":12},
-    {"name":"lint","passed":14,"failed":0,"total":14},
-    {"name":"severity","passed":14,"failed":0,"total":14},
-    {"name":"offenders","passed":19,"failed":0,"total":19},
-    {"name":"quorum","passed":9,"failed":0,"total":9},
-    {"name":"trace","passed":15,"failed":0,"total":15},
-    {"name":"whatif","passed":13,"failed":0,"total":13},
-    {"name":"batch","passed":17,"failed":0,"total":17},
-    {"name":"temporal","passed":15,"failed":0,"total":15},
-    {"name":"sla","passed":15,"failed":0,"total":15},
-    {"name":"wire","passed":16,"failed":0,"total":16},
-    {"name":"disjunction","passed":10,"failed":0,"total":10},
-    {"name":"activity","passed":17,"failed":0,"total":17},
-    {"name":"policies","passed":14,"failed":0,"total":14},
-    {"name":"defrule","passed":11,"failed":0,"total":11},
-    {"name":"pipeline","passed":15,"failed":0,"total":15}
-  ],
-  "generated": "2026-06-06T19:40:03+00:00"
-}

lib/mod/scoreboard.md

@@ -1,27 +0,0 @@
-# mod scoreboard
-
-**390 / 390 passing** (0 failure(s)).
-
-| Suite | Passed | Total | Status |
-|-------|--------|-------|--------|
-| decide | 31 | 31 | ok |
-| audit | 29 | 29 | ok |
-| escalation | 46 | 46 | ok |
-| fed | 26 | 26 | ok |
-| extensions | 32 | 32 | ok |
-| link | 12 | 12 | ok |
-| lint | 14 | 14 | ok |
-| severity | 14 | 14 | ok |
-| offenders | 19 | 19 | ok |
-| quorum | 9 | 9 | ok |
-| trace | 15 | 15 | ok |
-| whatif | 13 | 13 | ok |
-| batch | 17 | 17 | ok |
-| temporal | 15 | 15 | ok |
-| sla | 15 | 15 | ok |
-| wire | 16 | 16 | ok |
-| disjunction | 10 | 10 | ok |
-| activity | 17 | 17 | ok |
-| policies | 14 | 14 | ok |
-| defrule | 11 | 11 | ok |
-| pipeline | 15 | 15 | ok |

lib/mod/severity.sx

@@ -1,60 +0,0 @@
-;; lib/mod/severity.sx — "strictest-wins" decision strategy.
-;;
-;; The default engine resolves precedence by rule ORDER (first proven clause wins,
-;; via pl-query-one). Some policies instead want the HARSHEST applicable sanction
-;; regardless of order. mod/decide-strictest collects every rule that proves
-;; (pl-query-all) and picks the highest-severity action. Same decision shape as
-;; the engine, plus :strategy. Built over the engine's helpers; engine untouched.
-
-(define
-  mod/action-severity
-  (fn
-    (action)
-    (cond
-      ((= action "ban") 4)
-      ((= action "remove") 3)
-      ((= action "hide") 2)
-      ((= action "escalate") 1)
-      (true 0))))
-
-(define
-  mod/strictest-sol
-  (fn
-    (sols)
-    (reduce
-      (fn
-        (acc s)
-        (if
-          (nil? acc)
-          s
-          (if
-            (<
-              (mod/action-severity (dict-get acc "Action"))
-              (mod/action-severity (dict-get s "Action")))
-            s
-            acc)))
-      nil
-      sols)))
-
-(define
-  mod/decide-strictest
-  (fn
-    (r reports rules)
-    (let
-      ((count (mod/report-count (mod/report-about r) reports))
-        (kinds (mod/classify-keywords r))
-        (id (mod/report-id r)))
-      (let
-        ((program (mod/build-program r count rules)))
-        (let
-          ((db (pl-load program)))
-          (let
-            ((sols (pl-query-all db (str "policy_action(" id ", Action, Rule)"))))
-            (let
-              ((best (mod/strictest-sol sols)))
-              (if
-                (nil? best)
-                {:action "keep" :proof {:goals (list) :evidence kinds :conditions (list) :rule "none" :count count} :report-id id :rule "none" :strategy "strictest"}
-                (let
-                  ((rule (mod/find-rule rules (dict-get best "Rule"))))
-                  {:action (mod/rule-action rule) :proof {:goals (mod/proof-goals db id (mod/rule-when rule)) :evidence kinds :conditions (mod/rule-when rule) :rule (mod/rule-name rule) :count count} :report-id id :rule (mod/rule-name rule) :strategy "strictest"})))))))))

lib/mod/sla.sx

@@ -1,47 +0,0 @@
-;; lib/mod/sla.sx — service-level sweep over pending lifecycle cases.
-;;
-;; Composes the Phase-3 lifecycle with the Ext-12 time dimension: a case left in a
-;; pending state (open / triaged / appealed) past a deadline has breached SLA and
-;; should resurface. A timed-case pairs a case with the tick it entered its
-;; current state (the caller stamps this — the lifecycle stays timeless and pure).
-;; Terminal states (decided / final) never breach.
-
-(define mod/pending-states (list "open" "triaged" "appealed"))
-(define mod/pending-state? (fn (s) (mod/member? s mod/pending-states)))
-
-(define mod/mk-timed-case (fn (c entered-at) {:entered-at entered-at :case c}))
-(define mod/tc-case (fn (tc) (get tc :case)))
-(define mod/tc-entered-at (fn (tc) (get tc :entered-at)))
-
-(define
-  mod/overdue?
-  (fn
-    (tc now deadline)
-    (if
-      (mod/pending-state? (mod/case-state (mod/tc-case tc)))
-      (< deadline (- now (mod/tc-entered-at tc)))
-      false)))
-
-(define
-  mod/sla-sweep
-  (fn
-    (timed-cases now deadline)
-    (reduce
-      (fn
-        (acc tc)
-        (if
-          (mod/overdue? tc now deadline)
-          (append
-            acc
-            (list (mod/report-id (mod/case-report (mod/tc-case tc)))))
-          acc))
-      (list)
-      timed-cases)))
-
-(define
-  mod/overdue-count
-  (fn
-    (timed-cases now deadline)
-    (len (mod/sla-sweep timed-cases now deadline))))
-
-(define mod/age (fn (tc now) (- now (mod/tc-entered-at tc))))

lib/mod/temporal.sx

@@ -1,62 +0,0 @@
-;; lib/mod/temporal.sx — burst detection over a time window.
-;;
-;; A plain report count can't tell a burst (N reports in minutes) from slow
-;; accumulation (N reports over months). mod/decide-temporal takes a `now` tick
-;; and a `window`, counts reports about the subject with :at within [now-window,
-;; now], asserts it as burst_count/2, and lets a `(:burst-at-least K)` rule fire
-;; only on a genuine burst. Time is supplied (deterministic), never clock-read.
-
-(define
-  mod/window-count
-  (fn
-    (subject reports now window)
-    (reduce
-      (fn
-        (acc r)
-        (if
-          (if
-            (= (mod/report-about r) subject)
-            (<= (- now window) (mod/report-at r))
-            false)
-          (+ acc 1)
-          acc))
-      0
-      reports)))
-
-(define
-  mod/build-temporal-program
-  (fn
-    (r count bcount rules)
-    (str
-      (mod/report-facts r count)
-      "\n"
-      "burst_count("
-      (mod/pl-quote (mod/report-about r))
-      ", "
-      bcount
-      ").\n"
-      (mod/rules->program rules))))
-
-(define
-  mod/decide-temporal
-  (fn
-    (r reports rules now window)
-    (let
-      ((about (mod/report-about r))
-        (id (mod/report-id r))
-        (kinds (mod/classify-keywords r)))
-      (let
-        ((count (mod/report-count about reports))
-          (bcount (mod/window-count about reports now window)))
-        (let
-          ((program (mod/build-temporal-program r count bcount rules)))
-          (let
-            ((db (pl-load program)))
-            (let
-              ((sol (pl-query-one db (str "policy_action(" id ", Action, Rule)"))))
-              (if
-                (nil? sol)
-                {:action "keep" :proof {:burst bcount :goals (list) :evidence kinds :conditions (list) :rule "none" :count count} :report-id id :rule "none" :strategy "temporal"}
-                (let
-                  ((rule (mod/find-rule rules (dict-get sol "Rule"))))
-                  {:action (mod/rule-action rule) :proof {:burst bcount :goals (mod/proof-goals db id (mod/rule-when rule)) :evidence kinds :conditions (mod/rule-when rule) :rule (mod/rule-name rule) :count count} :report-id id :rule (mod/rule-name rule) :strategy "temporal"})))))))))

lib/mod/tests/activity.sx

@@ -1,95 +0,0 @@
-;; lib/mod/tests/activity.sx — Ext 16: ActivityPub-shaped decision export.
-
-(define mod-ap-count 0)
-(define mod-ap-pass 0)
-(define mod-ap-fail 0)
-(define mod-ap-failures (list))
-
-(define
-  mod-ap-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-ap-count (+ mod-ap-count 1))
-      (if
-        (= got expected)
-        (set! mod-ap-pass (+ mod-ap-pass 1))
-        (begin
-          (set! mod-ap-fail (+ mod-ap-fail 1))
-          (append!
-            mod-ap-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── action → AP verb ──
-
-(mod-ap-test! "remove → Delete" (mod/action->verb "remove") "Delete")
-(mod-ap-test! "ban → Block" (mod/action->verb "ban") "Block")
-(mod-ap-test! "hide → Flag" (mod/action->verb "hide") "Flag")
-(mod-ap-test! "escalate → Flag" (mod/action->verb "escalate") "Flag")
-(mod-ap-test! "keep → nil (no activity)" (mod/action->verb "keep") nil)
-
-;; ── single decision → activity ──
-
-(define mod-ap-spam (mod/mk-report "r1" "a" "bob" "this is spam"))
-(define
-  mod-ap-dec
-  (mod/decide-report mod-ap-spam (list mod-ap-spam) mod/default-rules))
-(define mod-ap-act (mod/decision->activity mod-ap-dec "instance.example"))
-
-(mod-ap-test! "activity type is Flag (hide)" (get mod-ap-act :type) "Flag")
-(mod-ap-test! "activity object is report id" (get mod-ap-act :object) "r1")
-(mod-ap-test!
-  "activity actor preserved"
-  (get mod-ap-act :actor)
-  "instance.example")
-(mod-ap-test!
-  "activity preserves precise action"
-  (get mod-ap-act :action)
-  "hide")
-(mod-ap-test! "activity carries rule" (get mod-ap-act :rule) "spam-hide")
-(mod-ap-test!
-  "activity summary"
-  (get mod-ap-act :summary)
-  "moderation/hide via spam-hide")
-
-;; ── keep produces no activity ──
-
-(define mod-ap-clean (mod/mk-report "r2" "a" "b" "a fine post"))
-(define
-  mod-ap-keep
-  (mod/decide-report mod-ap-clean (list mod-ap-clean) mod/default-rules))
-(mod-ap-test!
-  "keep decision → nil activity"
-  (mod/decision->activity mod-ap-keep "x")
-  nil)
-
-;; ── abuse → Delete ──
-
-(define mod-ap-abuse (mod/mk-report "r3" "a" "b" "harassment here"))
-(define
-  mod-ap-abuse-dec
-  (mod/decide-report mod-ap-abuse (list mod-ap-abuse) mod/default-rules))
-(mod-ap-test!
-  "abuse decision → Delete activity"
-  (get (mod/decision->activity mod-ap-abuse-dec "x") :type)
-  "Delete")
-
-;; ── batch export drops keeps ──
-
-(define mod-ap-decisions (list mod-ap-dec mod-ap-keep mod-ap-abuse-dec))
-(define mod-ap-acts (mod/decisions->activities mod-ap-decisions "inst"))
-(mod-ap-test! "batch export drops the keep" (len mod-ap-acts) 2)
-(mod-ap-test!
-  "batch export first is the Flag"
-  (get (first mod-ap-acts) :type)
-  "Flag")
-(mod-ap-test!
-  "batch export second is the Delete"
-  (get (nth mod-ap-acts 1) :type)
-  "Delete")
-(mod-ap-test!
-  "empty decisions → no activities"
-  (mod/decisions->activities (list) "inst")
-  (list))
-
-(define mod-activity-tests-run! (fn () {:failures mod-ap-failures :total mod-ap-count :passed mod-ap-pass :failed mod-ap-fail}))

lib/mod/tests/audit.sx

@@ -1,187 +0,0 @@
-;; lib/mod/tests/audit.sx — Phase 2: evidence accumulation + proof tree + audit.
-
-(define mod-aud-count 0)
-(define mod-aud-pass 0)
-(define mod-aud-fail 0)
-(define mod-aud-failures (list))
-
-(define
-  mod-aud-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-aud-count (+ mod-aud-count 1))
-      (if
-        (= got expected)
-        (set! mod-aud-pass (+ mod-aud-pass 1))
-        (begin
-          (set! mod-aud-fail (+ mod-aud-fail 1))
-          (append!
-            mod-aud-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(define
-  mod-aud-decide1
-  (fn (r) (mod/decide-report r (list r) mod/default-rules)))
-
-;; ── proof tree: keyword classification ──
-
-(define
-  mod-aud-spam
-  (mod-aud-decide1 (mod/mk-report "r1" "alice" "bob" "this is spam")))
-(define mod-aud-spam-goals (get (get mod-aud-spam :proof) :goals))
-
-(mod-aud-test! "spam proof has one goal" (len mod-aud-spam-goals) 1)
-(mod-aud-test!
-  "spam proof goal text"
-  (get (first mod-aud-spam-goals) :goal)
-  "classification(r1, spam)")
-(mod-aud-test!
-  "spam proof goal solved"
-  (get (first mod-aud-spam-goals) :solved)
-  true)
-
-;; ── proof tree: count rule with real bindings ──
-
-(define mod-aud-rep-r (mod/mk-report "r3" "ann" "dave" "x"))
-(define
-  mod-aud-rep
-  (mod/decide-report
-    mod-aud-rep-r
-    (list mod-aud-rep-r mod-aud-rep-r mod-aud-rep-r)
-    mod/default-rules))
-(define mod-aud-rep-goals (get (get mod-aud-rep :proof) :goals))
-(define mod-aud-rep-binds (get (first mod-aud-rep-goals) :bindings))
-
-(mod-aud-test!
-  "count proof goal solved"
-  (get (first mod-aud-rep-goals) :solved)
-  true)
-(mod-aud-test! "count proof binding N" (dict-get mod-aud-rep-binds "N") "3")
-(mod-aud-test!
-  "count proof binding S (subject)"
-  (dict-get mod-aud-rep-binds "S")
-  "dave")
-
-;; ── proof tree: default keep has a 'true' goal ──
-
-(define
-  mod-aud-keep
-  (mod-aud-decide1 (mod/mk-report "rk" "a" "b" "a fine post")))
-(define mod-aud-keep-goals (get (get mod-aud-keep :proof) :goals))
-
-(mod-aud-test!
-  "keep proof goal text true"
-  (get (first mod-aud-keep-goals) :goal)
-  "true")
-(mod-aud-test!
-  "keep proof goal solved"
-  (get (first mod-aud-keep-goals) :solved)
-  true)
-
-;; ── evidence accumulation drives a rule ──
-
-(define
-  mod-aud-rev-r
-  (mod/attach-evidence
-    (mod/mk-report "re" "a" "carol" "neutral")
-    (mod/mk-evidence "confirmed-abuse" "human")))
-(define mod-aud-rev (mod-aud-decide1 mod-aud-rev-r))
-
-(mod-aud-test!
-  "evidence has length 1"
-  (len (mod/report-evidence mod-aud-rev-r))
-  1)
-(mod-aud-test!
-  "evidence reviewer-remove → remove"
-  (get mod-aud-rev :action)
-  "remove")
-(mod-aud-test!
-  "evidence reviewer-remove rule"
-  (get mod-aud-rev :rule)
-  "reviewer-remove")
-(mod-aud-test!
-  "evidence proof goal solved"
-  (get (first (get (get mod-aud-rev :proof) :goals)) :solved)
-  true)
-(mod-aud-test!
-  "no evidence → not reviewer-remove"
-  (get (mod-aud-decide1 (mod/mk-report "rn" "a" "b" "neutral")) :rule)
-  "default-keep")
-
-;; ── append-only audit log via the api ──
-
-(mod/reset!)
-(mod/report "alice" "bob" "this is spam")
-(mod/report "carol" "eve" "fine post")
-(define mod-aud-d1 (mod/decide "r1"))
-(define mod-aud-d2 (mod/decide "r2"))
-
-(mod-aud-test! "two decisions logged" (mod/audit-count) 2)
-(mod-aud-test!
-  "first entry seq 1"
-  (get (first (mod/audit-all)) :seq)
-  1)
-(mod-aud-test!
-  "audit r1 returns one entry"
-  (len (mod/audit "r1"))
-  1)
-(mod-aud-test!
-  "audit r1 action matches decision"
-  (get (first (mod/audit "r1")) :action)
-  (get mod-aud-d1 :action))
-(mod-aud-test!
-  "audit r1 rule matches decision"
-  (get (first (mod/audit "r1")) :rule)
-  "spam-hide")
-(mod-aud-test!
-  "audit r1 entry carries proof goals"
-  (len (get (get (first (mod/audit "r1")) :proof) :goals))
-  1)
-(mod-aud-test!
-  "audit r2 keep"
-  (get (first (mod/audit "r2")) :action)
-  "keep")
-(mod-aud-test! "audit unknown report → empty" (mod/audit "r99") (list))
-
-;; ── append-only: re-deciding appends, never mutates ──
-
-(define mod-aud-d1b (mod/decide "r1"))
-
-(mod-aud-test! "re-decide appends (count 3)" (mod/audit-count) 3)
-(mod-aud-test!
-  "audit r1 now has 2 entries"
-  (len (mod/audit "r1"))
-  2)
-(mod-aud-test!
-  "audit r1 seqs monotonic"
-  (get (nth (mod/audit "r1") 1) :seq)
-  3)
-(mod-aud-test!
-  "audit-latest r1 is seq 3"
-  (get (mod/audit-latest "r1") :seq)
-  3)
-(mod-aud-test!
-  "first r1 entry unchanged (still seq 1)"
-  (get (first (mod/audit "r1")) :seq)
-  1)
-
-;; ── evidence snapshot captured at decision time ──
-
-(mod/add-evidence "r2" "confirmed-abuse" "human")
-(define mod-aud-d2b (mod/decide "r2"))
-
-(mod-aud-test!
-  "post-evidence decision flips to remove"
-  (get mod-aud-d2b :action)
-  "remove")
-(mod-aud-test!
-  "audit snapshot records evidence kind"
-  (mod/evidence-kind (first (get (mod/audit-latest "r2") :evidence)))
-  "confirmed-abuse")
-(mod-aud-test!
-  "earlier r2 entry had empty evidence snapshot"
-  (len (get (first (mod/audit "r2")) :evidence))
-  0)
-
-(define mod-audit-tests-run! (fn () {:failures mod-aud-failures :total mod-aud-count :passed mod-aud-pass :failed mod-aud-fail}))

lib/mod/tests/batch.sx

@@ -1,101 +0,0 @@
-;; lib/mod/tests/batch.sx — Ext 11: batch triage + corpus analytics.
-
-(define mod-b-count 0)
-(define mod-b-pass 0)
-(define mod-b-fail 0)
-(define mod-b-failures (list))
-
-(define
-  mod-b-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-b-count (+ mod-b-count 1))
-      (if
-        (= got expected)
-        (set! mod-b-pass (+ mod-b-pass 1))
-        (begin
-          (set! mod-b-fail (+ mod-b-fail 1))
-          (append!
-            mod-b-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; corpus: 2 spam, 1 abuse, 2 clean — distinct subjects so the count rule stays quiet
-(define
-  mod-b-corpus
-  (list
-    (mod/mk-report "r1" "u" "s1" "this is spam")
-    (mod/mk-report "r2" "u" "s2" "buy now offer")
-    (mod/mk-report "r3" "u" "s3" "harassment here")
-    (mod/mk-report "r4" "u" "s4" "a fine post")
-    (mod/mk-report "r5" "u" "s5" "thanks for sharing")))
-
-(define mod-b-decisions (mod/decide-batch mod-b-corpus mod/default-rules))
-
-;; ── decide-batch ──
-
-(mod-b-test! "one decision per report" (len mod-b-decisions) 5)
-(mod-b-test!
-  "first decision is hide"
-  (get (first mod-b-decisions) :action)
-  "hide")
-
-;; ── action histogram ──
-
-(define mod-b-hist (mod/action-histogram mod-b-decisions))
-(mod-b-test! "histogram hide count" (get mod-b-hist :hide) 2)
-(mod-b-test! "histogram remove count" (get mod-b-hist :remove) 1)
-(mod-b-test! "histogram keep count" (get mod-b-hist :keep) 2)
-(mod-b-test! "histogram escalate count" (get mod-b-hist :escalate) 0)
-(mod-b-test! "histogram ban count" (get mod-b-hist :ban) 0)
-(mod-b-test!
-  "histogram totals match corpus"
-  (+
-    (+ (get mod-b-hist :hide) (get mod-b-hist :remove))
-    (+
-      (get mod-b-hist :keep)
-      (+ (get mod-b-hist :escalate) (get mod-b-hist :ban))))
-  5)
-
-;; ── rule coverage (empirical) ──
-
-(define mod-b-cov (mod/rule-coverage mod-b-corpus mod/default-rules))
-(mod-b-test! "coverage has one row per rule" (len mod-b-cov) 6)
-(mod-b-test!
-  "spam-hide fired twice"
-  (mod/rule-fire-count mod-b-decisions "spam-hide")
-  2)
-(mod-b-test!
-  "abuse-remove fired once"
-  (mod/rule-fire-count mod-b-decisions "abuse-remove")
-  1)
-(mod-b-test!
-  "default-keep fired twice"
-  (mod/rule-fire-count mod-b-decisions "default-keep")
-  2)
-
-;; ── never-fired: rules not exercised by this corpus ──
-
-(define mod-b-never (mod/never-fired mod-b-corpus mod/default-rules))
-(mod-b-test!
-  "exonerated-keep never fired"
-  (mod/member? "exonerated-keep" mod-b-never)
-  true)
-(mod-b-test!
-  "reviewer-remove never fired"
-  (mod/member? "reviewer-remove" mod-b-never)
-  true)
-(mod-b-test!
-  "repeated-escalate never fired"
-  (mod/member? "repeated-escalate" mod-b-never)
-  true)
-(mod-b-test!
-  "spam-hide DID fire (not in never-fired)"
-  (mod/member? "spam-hide" mod-b-never)
-  false)
-(mod-b-test!
-  "three rules never fired on this corpus"
-  (len mod-b-never)
-  3)
-
-(define mod-batch-tests-run! (fn () {:failures mod-b-failures :total mod-b-count :passed mod-b-pass :failed mod-b-fail}))

lib/mod/tests/decide.sx

@@ -1,215 +0,0 @@
-;; lib/mod/tests/decide.sx — Phase 1: report representation + simple policy.
-
-(define mod-dec-count 0)
-(define mod-dec-pass 0)
-(define mod-dec-fail 0)
-(define mod-dec-failures (list))
-
-(define
-  mod-dec-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-dec-count (+ mod-dec-count 1))
-      (if
-        (= got expected)
-        (set! mod-dec-pass (+ mod-dec-pass 1))
-        (begin
-          (set! mod-dec-fail (+ mod-dec-fail 1))
-          (append!
-            mod-dec-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; decide a single report (count over a 1-element registry)
-(define
-  mod-dec-one
-  (fn
-    (reason)
-    (let
-      ((r (mod/mk-report "r1" "alice" "bob" reason)))
-      (mod/decide-report r (list r) mod/default-rules))))
-
-(define mod-dec-action (fn (reason) (get (mod-dec-one reason) :action)))
-
-;; ── spam keyword → :hide ──
-
-(mod-dec-test!
-  "spam keyword 'spam' → hide"
-  (mod-dec-action "this is spam")
-  "hide")
-(mod-dec-test!
-  "spam keyword 'buy now' → hide"
-  (mod-dec-action "buy now while stocks last")
-  "hide")
-(mod-dec-test!
-  "spam keyword case-insensitive 'CLICK HERE' → hide"
-  (mod-dec-action "CLICK HERE now")
-  "hide")
-(mod-dec-test!
-  "spam keyword 'free money' → hide"
-  (mod-dec-action "win free money fast")
-  "hide")
-
-;; ── abuse keyword → :remove ──
-
-(mod-dec-test!
-  "abuse keyword 'harassment' → remove"
-  (mod-dec-action "ongoing harassment of users")
-  "remove")
-(mod-dec-test!
-  "abuse keyword 'threat' → remove"
-  (mod-dec-action "this is a threat")
-  "remove")
-(mod-dec-test!
-  "abuse keyword 'slur' → remove"
-  (mod-dec-action "contains a slur")
-  "remove")
-
-;; ── no rule → :keep ──
-
-(mod-dec-test!
-  "neutral reason → keep"
-  (mod-dec-action "I disagree with this post")
-  "keep")
-(mod-dec-test! "empty reason → keep" (mod-dec-action "") "keep")
-
-;; ── decision carries the matching rule (proof, not bare keyword) ──
-
-(mod-dec-test!
-  "spam decision rule name"
-  (get (mod-dec-one "this is spam") :rule)
-  "spam-hide")
-(mod-dec-test!
-  "keep decision rule name"
-  (get (mod-dec-one "fine post") :rule)
-  "default-keep")
-(mod-dec-test!
-  "abuse decision rule name"
-  (get (mod-dec-one "harassment here") :rule)
-  "abuse-remove")
-(mod-dec-test!
-  "spam proof :rule"
-  (get (get (mod-dec-one "spam!") :proof) :rule)
-  "spam-hide")
-(mod-dec-test!
-  "spam proof :evidence"
-  (get (get (mod-dec-one "spam!") :proof) :evidence)
-  (list "spam"))
-(mod-dec-test!
-  "spam proof :count"
-  (get (get (mod-dec-one "spam!") :proof) :count)
-  1)
-
-;; ── classification (evidence derivation) ──
-
-(mod-dec-test!
-  "classify spam"
-  (mod/classify-keywords (mod/mk-report "r1" "a" "b" "spam!"))
-  (list "spam"))
-(mod-dec-test!
-  "classify abuse"
-  (mod/classify-keywords (mod/mk-report "r1" "a" "b" "abuse"))
-  (list "abuse"))
-(mod-dec-test!
-  "classify neutral → empty"
-  (mod/classify-keywords (mod/mk-report "r1" "a" "b" "hello"))
-  (list))
-(mod-dec-test!
-  "classify both spam+abuse"
-  (mod/classify-keywords (mod/mk-report "r1" "a" "b" "spam and abuse"))
-  (list "spam" "abuse"))
-
-;; ── report-count + repeated → :escalate ──
-
-(define
-  mod-dec-three
-  (list
-    (mod/mk-report "r1" "a" "bob" "x")
-    (mod/mk-report "r2" "c" "bob" "y")
-    (mod/mk-report "r3" "d" "bob" "z")))
-
-(mod-dec-test!
-  "report-count counts subject"
-  (mod/report-count "bob" mod-dec-three)
-  3)
-(mod-dec-test!
-  "3 reports about subject → escalate"
-  (get
-    (mod/decide-report (first mod-dec-three) mod-dec-three mod/default-rules)
-    :action)
-  "escalate")
-(mod-dec-test!
-  "escalate rule name"
-  (get
-    (mod/decide-report (first mod-dec-three) mod-dec-three mod/default-rules)
-    :rule)
-  "repeated-escalate")
-
-(define
-  mod-dec-two
-  (list
-    (mod/mk-report "r1" "a" "carol" "x")
-    (mod/mk-report "r2" "c" "carol" "y")))
-
-(mod-dec-test!
-  "2 reports about subject → keep (below threshold)"
-  (get
-    (mod/decide-report (first mod-dec-two) mod-dec-two mod/default-rules)
-    :action)
-  "keep")
-
-;; ── precedence: spam beats repeated ──
-
-(define
-  mod-dec-spam-among-many
-  (list
-    (mod/mk-report "r1" "a" "dave" "buy now spam")
-    (mod/mk-report "r2" "c" "dave" "y")
-    (mod/mk-report "r3" "d" "dave" "z")))
-
-(mod-dec-test!
-  "spam wins over repeated (precedence)"
-  (get
-    (mod/decide-report
-      (first mod-dec-spam-among-many)
-      mod-dec-spam-among-many
-      mod/default-rules)
-    :action)
-  "hide")
-
-;; ── accessors ──
-
-(mod-dec-test!
-  "report-about accessor"
-  (mod/report-about (mod/mk-report "r1" "a" "bob" "x"))
-  "bob")
-(mod-dec-test!
-  "report-by accessor"
-  (mod/report-by (mod/mk-report "r1" "alice" "bob" "x"))
-  "alice")
-
-;; ── api registry ──
-
-(mod/reset!)
-(define mod-dec-r1 (mod/report "alice" "bob" "this is spam"))
-(define mod-dec-r2 (mod/report "carol" "eve" "fine post"))
-
-(mod-dec-test!
-  "mod/report assigns sequential id r1"
-  (mod/report-id mod-dec-r1)
-  "r1")
-(mod-dec-test!
-  "mod/report assigns sequential id r2"
-  (mod/report-id mod-dec-r2)
-  "r2")
-(mod-dec-test!
-  "mod/decide via registry → hide"
-  (get (mod/decide "r1") :action)
-  "hide")
-(mod-dec-test!
-  "mod/decide via registry → keep"
-  (get (mod/decide "r2") :action)
-  "keep")
-(mod-dec-test! "mod/decide unknown id → nil" (mod/decide "r99") nil)
-
-(define mod-decide-tests-run! (fn () {:failures mod-dec-failures :total mod-dec-count :passed mod-dec-pass :failed mod-dec-fail}))

lib/mod/tests/defrule.sx

@@ -1,95 +0,0 @@
-;; lib/mod/tests/defrule.sx — Ext 18: ergonomic defrule / ruleset.
-
-(define mod-dr-count 0)
-(define mod-dr-pass 0)
-(define mod-dr-fail 0)
-(define mod-dr-failures (list))
-
-(define
-  mod-dr-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-dr-count (+ mod-dr-count 1))
-      (if
-        (= got expected)
-        (set! mod-dr-pass (+ mod-dr-pass 1))
-        (begin
-          (set! mod-dr-fail (+ mod-dr-fail 1))
-          (append!
-            mod-dr-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── defrule produces the same structure as mk-rule ──
-
-(define
-  mod-dr-r
-  (mod/defrule "spam-hide" :hide (list :classification "spam")))
-(mod-dr-test! "defrule name" (mod/rule-name mod-dr-r) "spam-hide")
-(mod-dr-test! "defrule action" (mod/rule-action mod-dr-r) "hide")
-(mod-dr-test!
-  "defrule when wraps the conditions"
-  (mod/rule-when mod-dr-r)
-  (list (list :classification "spam")))
-(mod-dr-test!
-  "defrule equals mk-rule equivalent"
-  (mod/rule-when mod-dr-r)
-  (mod/rule-when
-    (mod/mk-rule "spam-hide" :hide (list (list :classification "spam")))))
-
-;; ── multi-condition + no-condition ──
-
-(define
-  mod-dr-multi
-  (mod/defrule
-    "strict"
-    :hide (list :classification "spam")
-    (list :not (list :attr "verified"))))
-(mod-dr-test!
-  "defrule collects multiple conditions"
-  (len (mod/rule-when mod-dr-multi))
-  2)
-
-(define mod-dr-catch (mod/defrule "default-keep" :keep))
-(mod-dr-test!
-  "defrule with no conditions is unconditional"
-  (mod/rule-when mod-dr-catch)
-  (list))
-
-;; ── ruleset assembles a list ──
-
-(define
-  mod-dr-rules
-  (mod/ruleset
-    (mod/defrule "spam-hide" :hide (list :classification "spam"))
-    (mod/defrule "default-keep" :keep)))
-
-(mod-dr-test! "ruleset length" (len mod-dr-rules) 2)
-(mod-dr-test!
-  "ruleset first rule name"
-  (mod/rule-name (first mod-dr-rules))
-  "spam-hide")
-
-;; ── engine works with defrule/ruleset-built policy ──
-
-(define mod-dr-spam (mod/mk-report "r1" "a" "b" "this is spam"))
-(define mod-dr-clean (mod/mk-report "r2" "a" "b" "a fine post"))
-
-(mod-dr-test!
-  "defrule policy: spam → hide"
-  (get
-    (mod/decide-report mod-dr-spam (list mod-dr-spam) mod-dr-rules)
-    :action)
-  "hide")
-(mod-dr-test!
-  "defrule policy: clean → keep"
-  (get
-    (mod/decide-report mod-dr-clean (list mod-dr-clean) mod-dr-rules)
-    :action)
-  "keep")
-(mod-dr-test!
-  "defrule policy: spam names the rule"
-  (get (mod/decide-report mod-dr-spam (list mod-dr-spam) mod-dr-rules) :rule)
-  "spam-hide")
-
-(define mod-defrule-tests-run! (fn () {:failures mod-dr-failures :total mod-dr-count :passed mod-dr-pass :failed mod-dr-fail}))

lib/mod/tests/disjunction.sx

@@ -1,145 +0,0 @@
-;; lib/mod/tests/disjunction.sx — Ext 15: disjunctive (:any) conditions.
-
-(define mod-or-count 0)
-(define mod-or-pass 0)
-(define mod-or-fail 0)
-(define mod-or-failures (list))
-
-(define
-  mod-or-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-or-count (+ mod-or-count 1))
-      (if
-        (= got expected)
-        (set! mod-or-pass (+ mod-or-pass 1))
-        (begin
-          (set! mod-or-fail (+ mod-or-fail 1))
-          (append!
-            mod-or-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; one rule, OR of two classifications → one action covers both
-(define
-  mod-or-rules
-  (list
-    (mod/mk-rule
-      "spam-or-abuse-hide"
-      :hide (list
-        (list
-          :any (list (list :classification "spam") (list :classification "abuse")))))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define mod-or-spam (mod/mk-report "r1" "a" "b" "this is spam"))
-(define mod-or-abuse (mod/mk-report "r2" "a" "b" "harassment here"))
-(define mod-or-clean (mod/mk-report "r3" "a" "b" "a fine post"))
-
-(mod-or-test!
-  "OR: spam branch → hide"
-  (get
-    (mod/decide-report mod-or-spam (list mod-or-spam) mod-or-rules)
-    :action)
-  "hide")
-(mod-or-test!
-  "OR: abuse branch → hide"
-  (get
-    (mod/decide-report mod-or-abuse (list mod-or-abuse) mod-or-rules)
-    :action)
-  "hide")
-(mod-or-test!
-  "OR: neither branch → keep"
-  (get
-    (mod/decide-report mod-or-clean (list mod-or-clean) mod-or-rules)
-    :action)
-  "keep")
-
-;; ── goal text + proof ──
-
-(mod-or-test!
-  "cond->goal :any joins with ;"
-  (mod/cond->goal
-    (list
-      :any (list (list :classification "spam") (list :classification "abuse")))
-    "Id")
-  "(classification(Id, spam) ; classification(Id, abuse))")
-
-(define
-  mod-or-dec
-  (mod/decide-report mod-or-spam (list mod-or-spam) mod-or-rules))
-(mod-or-test!
-  "OR proof goal solved"
-  (get (first (get (get mod-or-dec :proof) :goals)) :solved)
-  true)
-(mod-or-test!
-  "OR proof goal text"
-  (get (first (get (get mod-or-dec :proof) :goals)) :goal)
-  "(classification(r1, spam) ; classification(r1, abuse))")
-
-;; ── :any composes with :not (NOR-ish) and :attr ──
-
-(define
-  mod-or-mixed-rules
-  (list
-    (mod/mk-rule
-      "spam-or-flagged-hide"
-      :hide (list
-        (list
-          :any (list (list :classification "spam") (list :attr "flagged")))))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define
-  mod-or-flagged
-  (mod/attach-attr (mod/mk-report "r4" "a" "b" "a fine post") "flagged"))
-(mod-or-test!
-  "OR over classification|attr: flagged clean post → hide"
-  (get
-    (mod/decide-report
-      mod-or-flagged
-      (list mod-or-flagged)
-      mod-or-mixed-rules)
-    :action)
-  "hide")
-
-(mod-or-test!
-  "cond->goal :any with :not branch"
-  (mod/cond->goal
-    (list
-      :any (list
-        (list :classification "spam")
-        (list :not (list :attr "verified"))))
-    "Id")
-  "(classification(Id, spam) ; not(attr(Id, verified)))")
-
-;; AND still works alongside OR in the same :when list
-(define
-  mod-or-and-rules
-  (list
-    (mod/mk-rule
-      "spam-and-not-verified"
-      :hide (list
-        (list
-          :any (list (list :classification "spam") (list :classification "abuse")))
-        (list :not (list :attr "verified"))))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define
-  mod-or-spam-verified
-  (mod/attach-attr (mod/mk-report "r5" "a" "b" "this is spam") "verified"))
-(mod-or-test!
-  "AND of OR + NOT: verified spam → keep"
-  (get
-    (mod/decide-report
-      mod-or-spam-verified
-      (list mod-or-spam-verified)
-      mod-or-and-rules)
-    :action)
-  "keep")
-(mod-or-test!
-  "AND of OR + NOT: unverified abuse → hide"
-  (get
-    (mod/decide-report mod-or-abuse (list mod-or-abuse) mod-or-and-rules)
-    :action)
-  "hide")
-
-(define mod-disjunction-tests-run! (fn () {:failures mod-or-failures :total mod-or-count :passed mod-or-pass :failed mod-or-fail}))

lib/mod/tests/escalation.sx

@@ -1,279 +0,0 @@
-;; lib/mod/tests/escalation.sx — Phase 3: lifecycle state machine + escalation.
-
-(define mod-esc-count 0)
-(define mod-esc-pass 0)
-(define mod-esc-fail 0)
-(define mod-esc-failures (list))
-
-(define
-  mod-esc-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-esc-count (+ mod-esc-count 1))
-      (if
-        (= got expected)
-        (set! mod-esc-pass (+ mod-esc-pass 1))
-        (begin
-          (set! mod-esc-fail (+ mod-esc-fail 1))
-          (append!
-            mod-esc-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── transition table guard ──
-
-(mod-esc-test!
-  "open → triaged allowed"
-  (mod/lc-can-transition? "open" "triaged")
-  true)
-(mod-esc-test!
-  "triaged → decided allowed"
-  (mod/lc-can-transition? "triaged" "decided")
-  true)
-(mod-esc-test!
-  "decided → appealed allowed"
-  (mod/lc-can-transition? "decided" "appealed")
-  true)
-(mod-esc-test!
-  "appealed → final allowed"
-  (mod/lc-can-transition? "appealed" "final")
-  true)
-(mod-esc-test!
-  "open → decided rejected"
-  (mod/lc-can-transition? "open" "decided")
-  false)
-(mod-esc-test!
-  "triaged → final rejected"
-  (mod/lc-can-transition? "triaged" "final")
-  false)
-(mod-esc-test!
-  "final is terminal"
-  (mod/lc-can-transition? "final" "open")
-  false)
-
-;; ── initial state ──
-
-(define
-  mod-esc-c0
-  (mod/mk-case (mod/mk-report "r1" "alice" "bob" "this is spam")))
-(mod-esc-test! "new case is open" (mod/case-state mod-esc-c0) "open")
-(mod-esc-test! "new case has no decision" (mod/case-decision mod-esc-c0) nil)
-
-;; ── auto-tier: spam triages + resolves to decided/hide ──
-
-(define
-  mod-esc-spam-rep
-  (list (mod/mk-report "r1" "alice" "bob" "this is spam")))
-(define
-  mod-esc-t1
-  (mod/case-triage mod-esc-c0 mod-esc-spam-rep mod/default-rules))
-(mod-esc-test! "spam triaged" (mod/case-state mod-esc-t1) "triaged")
-(mod-esc-test! "spam triage tier auto" (mod/case-tier mod-esc-t1) "auto")
-(mod-esc-test! "spam triage action hide" (mod/case-action mod-esc-t1) "hide")
-
-(define mod-esc-r1 (mod/case-resolve mod-esc-t1))
-(mod-esc-test!
-  "auto resolve → decided"
-  (mod/case-state mod-esc-r1)
-  "decided")
-(mod-esc-test!
-  "decision preserved through resolve"
-  (mod/case-action mod-esc-r1)
-  "hide")
-
-;; ── illegal transition flags :error, leaves state ──
-
-(define mod-esc-bad (mod/case-finalize mod-esc-c0))
-(mod-esc-test!
-  "finalize from open is illegal"
-  (mod/case-state mod-esc-bad)
-  "open")
-(mod-esc-test!
-  "illegal transition sets error"
-  (nil? (mod/case-error mod-esc-bad))
-  false)
-
-;; ── human-tier: repeated report escalates, resolve blocked, review decides ──
-
-(define mod-esc-rep-r (mod/mk-report "r3" "ann" "dave" "off-topic"))
-(define mod-esc-rep-reports (list mod-esc-rep-r mod-esc-rep-r mod-esc-rep-r))
-(define mod-esc-rep-c0 (mod/mk-case mod-esc-rep-r))
-(define
-  mod-esc-rep-t
-  (mod/case-triage mod-esc-rep-c0 mod-esc-rep-reports mod/default-rules))
-
-(mod-esc-test!
-  "repeated triage action escalate"
-  (mod/case-action mod-esc-rep-t)
-  "escalate")
-(mod-esc-test!
-  "repeated triage tier human"
-  (mod/case-tier mod-esc-rep-t)
-  "human")
-(mod-esc-test!
-  "repeated still triaged after triage"
-  (mod/case-state mod-esc-rep-t)
-  "triaged")
-
-(define mod-esc-rep-block (mod/case-resolve mod-esc-rep-t))
-(mod-esc-test!
-  "auto-resolve blocked on human tier (state unchanged)"
-  (mod/case-state mod-esc-rep-block)
-  "triaged")
-(mod-esc-test!
-  "blocked resolve sets error"
-  (nil? (mod/case-error mod-esc-rep-block))
-  false)
-
-(define
-  mod-esc-rep-rev
-  (mod/case-review
-    mod-esc-rep-t
-    "confirmed-abuse"
-    "human"
-    mod-esc-rep-reports
-    mod/default-rules))
-(mod-esc-test!
-  "human review → decided"
-  (mod/case-state mod-esc-rep-rev)
-  "decided")
-(mod-esc-test!
-  "human review action remove"
-  (mod/case-action mod-esc-rep-rev)
-  "remove")
-(mod-esc-test!
-  "review attached evidence to report"
-  (len (mod/report-evidence (mod/case-report mod-esc-rep-rev)))
-  1)
-
-(define mod-esc-rep-final (mod/case-finalize mod-esc-rep-rev))
-(mod-esc-test!
-  "review case finalizes"
-  (mod/case-state mod-esc-rep-final)
-  "final")
-
-;; ── appeal overrides a prior decision ──
-
-(define
-  mod-esc-ap-c0
-  (mod/mk-case (mod/mk-report "r5" "u" "v" "buy now spam")))
-(define mod-esc-ap-rep (list (mod/mk-report "r5" "u" "v" "buy now spam")))
-(define
-  mod-esc-ap-t
-  (mod/case-triage mod-esc-ap-c0 mod-esc-ap-rep mod/default-rules))
-(define mod-esc-ap-d (mod/case-resolve mod-esc-ap-t))
-
-(mod-esc-test!
-  "appeal precondition decided/hide"
-  (mod/case-action mod-esc-ap-d)
-  "hide")
-
-(define
-  mod-esc-ap-appealed
-  (mod/case-appeal
-    mod-esc-ap-d
-    "exonerated"
-    "moderator"
-    mod-esc-ap-rep
-    mod/default-rules))
-(mod-esc-test!
-  "appeal → appealed state"
-  (mod/case-state mod-esc-ap-appealed)
-  "appealed")
-(mod-esc-test!
-  "appeal overrides hide → keep"
-  (mod/case-action mod-esc-ap-appealed)
-  "keep")
-(mod-esc-test!
-  "appeal recorded via exonerated-keep rule"
-  (get (mod/case-decision mod-esc-ap-appealed) :rule)
-  "exonerated-keep")
-
-(define mod-esc-ap-final (mod/case-finalize mod-esc-ap-appealed))
-(mod-esc-test! "appealed → final" (mod/case-state mod-esc-ap-final) "final")
-
-;; ── history records the full traversal ──
-
-(mod-esc-test!
-  "full lifecycle history length 4 (triage,resolve,appeal,finalize)"
-  (len (mod/case-history mod-esc-ap-final))
-  4)
-(mod-esc-test!
-  "first history step open→triaged"
-  (get (first (mod/case-history mod-esc-ap-final)) :to)
-  "triaged")
-(mod-esc-test!
-  "last history step → final"
-  (get (nth (mod/case-history mod-esc-ap-final) 3) :to)
-  "final")
-
-;; ── api-level lifecycle façade ──
-
-(mod/reset!)
-(mod/report "alice" "bob" "this is spam")
-(mod/report "carol" "dave" "off-topic")
-(mod/report "carol" "dave" "off-topic")
-(mod/report "carol" "dave" "off-topic")
-
-(mod-esc-test!
-  "api: case opens at open"
-  (mod/case-state (mod/case-of "r1"))
-  "open")
-
-(define mod-esc-api-t1 (mod/triage "r1"))
-(mod-esc-test!
-  "api: triage spam → triaged"
-  (mod/case-state mod-esc-api-t1)
-  "triaged")
-(mod-esc-test!
-  "api: triage spam action hide"
-  (mod/case-action mod-esc-api-t1)
-  "hide")
-
-(define mod-esc-api-r1 (mod/resolve "r1"))
-(mod-esc-test!
-  "api: resolve → decided"
-  (mod/case-state mod-esc-api-r1)
-  "decided")
-(mod-esc-test!
-  "api: resolve logged decision"
-  (len (mod/audit "r1"))
-  1)
-
-(define mod-esc-api-app (mod/appeal "r1" "exonerated" "mod"))
-(mod-esc-test!
-  "api: appeal → appealed"
-  (mod/case-state mod-esc-api-app)
-  "appealed")
-(mod-esc-test!
-  "api: appeal overrides → keep"
-  (mod/case-action mod-esc-api-app)
-  "keep")
-(mod-esc-test!
-  "api: appeal logged second decision"
-  (len (mod/audit "r1"))
-  2)
-(mod-esc-test!
-  "api: finalize → final"
-  (mod/case-state (mod/finalize "r1"))
-  "final")
-
-;; r4 is the 3rd report about dave → escalates via the human tier
-(define mod-esc-api-t4 (mod/triage "r4"))
-(mod-esc-test!
-  "api: repeated triage escalates (human tier)"
-  (mod/case-tier mod-esc-api-t4)
-  "human")
-(define mod-esc-api-blk (mod/resolve "r4"))
-(mod-esc-test!
-  "api: escalated resolve blocked"
-  (mod/case-state mod-esc-api-blk)
-  "triaged")
-(define mod-esc-api-rev (mod/review "r4" "confirmed-abuse" "human"))
-(mod-esc-test!
-  "api: review → decided/remove"
-  (mod/case-action mod-esc-api-rev)
-  "remove")
-(mod-esc-test! "api: unknown id → nil" (mod/triage "r99") nil)
-
-(define mod-escalation-tests-run! (fn () {:failures mod-esc-failures :total mod-esc-count :passed mod-esc-pass :failed mod-esc-fail}))

lib/mod/tests/extensions.sx

@@ -1,313 +0,0 @@
-;; lib/mod/tests/extensions.sx — beyond-roadmap extensions.
-;;
-;; Ext 1: negation-as-failure conditions (:not / :attr) + report attributes.
-;;        "hide spam UNLESS the author is verified" (closed-world reasoning).
-;; Ext 2: weighted/aggregate evidence scoring (:score-at-least) + report signals.
-;;        Many low-confidence signals accumulate past a threshold via Prolog
-;;        aggregate_all(sum(W), ...).
-;; Ext 3: human-readable proof explanation (mod/explain) over the proof tree.
-;; Demonstrated with custom rule sets so the default policy (and its conformance
-;; tests) stays untouched.
-
-(define mod-ext-count 0)
-(define mod-ext-pass 0)
-(define mod-ext-fail 0)
-(define mod-ext-failures (list))
-
-(define
-  mod-ext-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-ext-count (+ mod-ext-count 1))
-      (if
-        (= got expected)
-        (set! mod-ext-pass (+ mod-ext-pass 1))
-        (begin
-          (set! mod-ext-fail (+ mod-ext-fail 1))
-          (append!
-            mod-ext-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── Ext 1: report attributes ──
-
-(define mod-ext-r0 (mod/mk-report "r1" "a" "b" "this is spam"))
-(mod-ext-test!
-  "fresh report has no attrs"
-  (len (mod/report-attrs mod-ext-r0))
-  0)
-(define mod-ext-rv (mod/attach-attr mod-ext-r0 "verified"))
-(mod-ext-test!
-  "attach-attr adds one attr"
-  (len (mod/report-attrs mod-ext-rv))
-  1)
-(mod-ext-test!
-  "attach-attr preserves evidence field"
-  (len
-    (mod/report-evidence
-      (mod/attach-evidence mod-ext-rv (mod/mk-evidence "x" "y"))))
-  1)
-(mod-ext-test!
-  "attach-evidence preserves attrs"
-  (len
-    (mod/report-attrs
-      (mod/attach-evidence mod-ext-rv (mod/mk-evidence "x" "y"))))
-  1)
-
-;; ── Ext 1: negation-as-failure: spam hidden unless author verified ──
-
-(define
-  mod-ext-rules
-  (list
-    (mod/mk-rule
-      "spam-unverified-hide"
-      :hide (list
-        (list :classification "spam")
-        (list :not (list :attr "verified"))))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define mod-ext-spam-plain (mod/mk-report "p1" "a" "b" "this is spam"))
-(define
-  mod-ext-spam-verified
-  (mod/attach-attr (mod/mk-report "p2" "a" "b" "this is spam") "verified"))
-(define mod-ext-clean (mod/mk-report "p3" "a" "b" "a fine post"))
-
-(mod-ext-test!
-  "unverified spam → hide"
-  (get
-    (mod/decide-report
-      mod-ext-spam-plain
-      (list mod-ext-spam-plain)
-      mod-ext-rules)
-    :action)
-  "hide")
-(mod-ext-test!
-  "verified author spam → keep (negation blocks)"
-  (get
-    (mod/decide-report
-      mod-ext-spam-verified
-      (list mod-ext-spam-verified)
-      mod-ext-rules)
-    :action)
-  "keep")
-(mod-ext-test!
-  "clean post → keep"
-  (get
-    (mod/decide-report mod-ext-clean (list mod-ext-clean) mod-ext-rules)
-    :action)
-  "keep")
-
-;; ── Ext 1: negation appears in the goal text + proof ──
-
-(define
-  mod-ext-dec
-  (mod/decide-report
-    mod-ext-spam-plain
-    (list mod-ext-spam-plain)
-    mod-ext-rules))
-(define mod-ext-goals (get (get mod-ext-dec :proof) :goals))
-
-(mod-ext-test!
-  "rule that matched is spam-unverified-hide"
-  (get mod-ext-dec :rule)
-  "spam-unverified-hide")
-(mod-ext-test! "proof has two goals" (len mod-ext-goals) 2)
-(mod-ext-test!
-  "negation goal text"
-  (get (nth mod-ext-goals 1) :goal)
-  "not(attr(p1, verified))")
-(mod-ext-test!
-  "negation goal solved for unverified"
-  (get (nth mod-ext-goals 1) :solved)
-  true)
-
-;; ── Ext 1: cond->goal compiles :attr and :not directly ──
-
-(mod-ext-test!
-  "cond->goal :attr"
-  (mod/cond->goal (list :attr "verified") "Id")
-  "attr(Id, verified)")
-(mod-ext-test!
-  "cond->goal :not wraps inner"
-  (mod/cond->goal (list :not (list :classification "spam")) "Id")
-  "not(classification(Id, spam))")
-
-;; ── Ext 1: positive :attr condition (allowlist-style) ──
-
-(define
-  mod-ext-allow-rules
-  (list
-    (mod/mk-rule "trusted-keep" :keep (list (list :attr "trusted")))
-    (mod/mk-rule "spam-hide" :hide (list (list :classification "spam")))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define
-  mod-ext-trusted-spam
-  (mod/attach-attr (mod/mk-report "t1" "a" "b" "this is spam") "trusted"))
-(mod-ext-test!
-  "trusted attr exempts spam → keep"
-  (get
-    (mod/decide-report
-      mod-ext-trusted-spam
-      (list mod-ext-trusted-spam)
-      mod-ext-allow-rules)
-    :action)
-  "keep")
-
-;; ── Ext 2: weighted signals + aggregate scoring ──
-
-(define mod-ext-s0 (mod/mk-report "s1" "a" "b" "neutral"))
-(mod-ext-test!
-  "fresh report has no signals"
-  (len (mod/report-signals mod-ext-s0))
-  0)
-(define
-  mod-ext-s1
-  (mod/attach-signal mod-ext-s0 (mod/mk-signal "link" 2)))
-(mod-ext-test!
-  "attach-signal adds one"
-  (len (mod/report-signals mod-ext-s1))
-  1)
-(mod-ext-test!
-  "attach-signal preserves attrs"
-  (len
-    (mod/report-attrs
-      (mod/attach-signal mod-ext-rv (mod/mk-signal "x" 1))))
-  1)
-
-(define
-  mod-ext-score-rules
-  (list
-    (mod/mk-rule
-      "high-score-hide"
-      :hide (list (list :score-at-least 5)))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-;; one weak signal (2) — below threshold
-(define
-  mod-ext-weak
-  (mod/attach-signal
-    (mod/mk-report "w1" "a" "b" "neutral")
-    (mod/mk-signal "link" 2)))
-(mod-ext-test!
-  "single weak signal → keep (below threshold)"
-  (get
-    (mod/decide-report mod-ext-weak (list mod-ext-weak) mod-ext-score-rules)
-    :action)
-  "keep")
-
-;; three signals summing to 6 — over threshold
-(define
-  mod-ext-strong0
-  (mod/attach-signal
-    (mod/mk-report "w2" "a" "b" "neutral")
-    (mod/mk-signal "link" 2)))
-(define
-  mod-ext-strong1
-  (mod/attach-signal mod-ext-strong0 (mod/mk-signal "newaccount" 2)))
-(define
-  mod-ext-strong
-  (mod/attach-signal mod-ext-strong1 (mod/mk-signal "burst" 2)))
-(mod-ext-test!
-  "accumulated signals (2+2+2=6) → hide"
-  (get
-    (mod/decide-report
-      mod-ext-strong
-      (list mod-ext-strong)
-      mod-ext-score-rules)
-    :action)
-  "hide")
-(mod-ext-test!
-  "scoring rule named in decision"
-  (get
-    (mod/decide-report
-      mod-ext-strong
-      (list mod-ext-strong)
-      mod-ext-score-rules)
-    :rule)
-  "high-score-hide")
-
-;; exactly at threshold (5) fires
-(define
-  mod-ext-exact0
-  (mod/attach-signal
-    (mod/mk-report "w3" "a" "b" "neutral")
-    (mod/mk-signal "link" 3)))
-(define
-  mod-ext-exact
-  (mod/attach-signal mod-ext-exact0 (mod/mk-signal "burst" 2)))
-(mod-ext-test!
-  "exactly at threshold (5) → hide"
-  (get
-    (mod/decide-report mod-ext-exact (list mod-ext-exact) mod-ext-score-rules)
-    :action)
-  "hide")
-
-(mod-ext-test!
-  "cond->goal :score-at-least"
-  (mod/cond->goal (list :score-at-least 5) "Id")
-  "aggregate_all(sum(W), signal(Id, _, W), T), T >= 5")
-
-;; ── Ext 3: human-readable proof explanation ──
-
-(define mod-ext-spam-explain (mod/explain mod-ext-dec))
-
-(mod-ext-test!
-  "explain mentions the report id"
-  (mod/str-contains? mod-ext-spam-explain "Report p1")
-  true)
-(mod-ext-test!
-  "explain mentions the action"
-  (mod/str-contains? mod-ext-spam-explain "hide")
-  true)
-(mod-ext-test!
-  "explain mentions the rule"
-  (mod/str-contains? mod-ext-spam-explain "spam-unverified-hide")
-  true)
-(mod-ext-test!
-  "explain marks proved goals"
-  (mod/str-contains? mod-ext-spam-explain "[proved]")
-  true)
-(mod-ext-test!
-  "explain renders the evidence line"
-  (mod/str-contains? mod-ext-spam-explain "Evidence: spam")
-  true)
-
-;; count-rule explanation shows the unification bindings
-(define mod-ext-rep-r (mod/mk-report "rc" "ann" "dave" "off-topic"))
-(define
-  mod-ext-rep-d
-  (mod/decide-report
-    mod-ext-rep-r
-    (list mod-ext-rep-r mod-ext-rep-r mod-ext-rep-r)
-    mod/default-rules))
-(define mod-ext-rep-explain (mod/explain mod-ext-rep-d))
-(mod-ext-test!
-  "explain shows binding N=3"
-  (mod/str-contains? mod-ext-rep-explain "N=3")
-  true)
-(mod-ext-test!
-  "explain shows subject binding"
-  (mod/str-contains? mod-ext-rep-explain "dave")
-  true)
-
-;; explain-goal direct: unproved goal gets [unproved]
-(mod-ext-test!
-  "explain-goal marks unproved"
-  (mod/str-contains? (mod/explain-goal {:solved false :goal "attr(x, foo)" :bindings {}}) "[unproved]")
-  true)
-;; explain-binds renders key=value pairs
-(mod-ext-test!
-  "explain-binds renders pair"
-  (mod/explain-binds {:N "3"})
-  "N=3")
-;; no-evidence decision says (none)
-(define
-  mod-ext-keep-d
-  (mod/decide-report mod-ext-clean (list mod-ext-clean) mod-ext-rules))
-(mod-ext-test!
-  "explain (none) for empty evidence"
-  (mod/str-contains? (mod/explain mod-ext-keep-d) "Evidence: (none)")
-  true)
-
-(define mod-extensions-tests-run! (fn () {:failures mod-ext-failures :total mod-ext-count :passed mod-ext-pass :failed mod-ext-fail}))

lib/mod/tests/fed.sx

@@ -1,154 +0,0 @@
-;; lib/mod/tests/fed.sx — Phase 4: federation (mock fed-sx).
-
-(define mod-fed-count 0)
-(define mod-fed-pass 0)
-(define mod-fed-fail 0)
-(define mod-fed-failures (list))
-
-(define
-  mod-fed-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-fed-count (+ mod-fed-count 1))
-      (if
-        (= got expected)
-        (set! mod-fed-pass (+ mod-fed-pass 1))
-        (begin
-          (set! mod-fed-fail (+ mod-fed-fail 1))
-          (append!
-            mod-fed-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(mod/reset!)
-(mod/fed-reset!)
-
-;; ── trust model (advisory by default) ──
-
-(mod-fed-test! "trust initially false" (mod/trusted? "peerA" :mod) false)
-(mod/grant-trust "peerA" :mod)
-(mod-fed-test! "trust after grant" (mod/trusted? "peerA" :mod) true)
-(mod-fed-test! "trust wrong scope" (mod/trusted? "peerA" :other) false)
-(mod-fed-test! "trust other peer" (mod/trusted? "peerB" :mod) false)
-(mod/revoke-trust "peerA" :mod)
-(mod-fed-test! "trust after revoke" (mod/trusted? "peerA" :mod) false)
-
-;; ── cross-instance reports ──
-
-(define
-  mod-fed-fr
-  (mod/fed-receive-report "peerB" "alice" "bob" "this is spam"))
-(mod-fed-test! "fed report assigned id r1" (mod/report-id mod-fed-fr) "r1")
-(mod-fed-test! "fed report origin is peer" (mod/report-origin "r1") "peerB")
-(define mod-fed-local (mod/report "carol" "dave" "fine post"))
-(mod-fed-test!
-  "local report origin is local"
-  (mod/report-origin (mod/report-id mod-fed-local))
-  "local")
-(mod-fed-test!
-  "engine decides fed report (spam → hide)"
-  (get
-    (mod/decide-report mod-fed-fr (list mod-fed-fr) mod/default-rules)
-    :action)
-  "hide")
-
-;; ── decision sharing (outbox) ──
-
-(define mod-fed-dec {:action "hide" :rule "spam-hide" :report-id "r1"})
-(define
-  mod-fed-shared
-  (mod/fed-share-decision mod-fed-dec (list "peerB" "peerC")))
-(mod-fed-test! "share returns notified peers" (len mod-fed-shared) 2)
-(mod-fed-test! "outbox has two messages" (len (mod/fed-outbox)) 2)
-(mod-fed-test!
-  "outbox message type decision"
-  (get (first (mod/fed-outbox)) :type)
-  "decision")
-(mod-fed-test!
-  "outbox message addressed to peer"
-  (get (first (mod/fed-outbox)) :to)
-  "peerB")
-
-;; ── receiving a peer decision: advisory unless trusted ──
-
-(define mod-fed-untrusted (mod/fed-receive-decision "peerZ" {:action "remove" :rule "reviewer-remove" :report-id "rx"}))
-(mod-fed-test!
-  "untrusted decision not applied"
-  (get mod-fed-untrusted :applied)
-  false)
-(mod-fed-test!
-  "untrusted decision advisory"
-  (get mod-fed-untrusted :advisory)
-  true)
-(mod-fed-test!
-  "untrusted decision absent from applied log"
-  (mod/fed-applied-action "rx")
-  nil)
-(mod-fed-test!
-  "advisory log records suggestion"
-  (len mod/*fed-advisory*)
-  1)
-
-(mod/grant-trust "peerT" :mod)
-(define mod-fed-trusted (mod/fed-receive-decision "peerT" {:action "hide" :rule "spam-hide" :report-id "ry"}))
-(mod-fed-test! "trusted decision applied" (get mod-fed-trusted :applied) true)
-(mod-fed-test!
-  "trusted decision binds locally"
-  (get (mod/fed-applied-action "ry") :action)
-  "hide")
-
-;; ── revocation ──
-
-(mod-fed-test!
-  "applied action not yet revoked"
-  (get (mod/fed-applied-action "ry") :revoked)
-  false)
-(mod/fed-revoke! "ry" "manual")
-(mod-fed-test!
-  "revoke marks applied action revoked"
-  (get (mod/fed-applied-action "ry") :revoked)
-  true)
-(mod-fed-test!
-  "revoke emits a revocation message"
-  (mod/any? (fn (m) (= (get m :type) "revocation")) (mod/fed-outbox))
-  true)
-
-;; revoke-if-invalidated: proof still holds → no revocation
-(define mod-fed-spam-r (mod/mk-report "rs" "a" "b" "this is spam"))
-(define
-  mod-fed-spam-d
-  (mod/decide-report mod-fed-spam-r (list mod-fed-spam-r) mod/default-rules))
-(mod-fed-test! "spam decision is hide" (get mod-fed-spam-d :action) "hide")
-(define
-  mod-fed-rev-same
-  (mod/fed-revoke-if-invalidated
-    mod-fed-spam-r
-    mod-fed-spam-d
-    (list mod-fed-spam-r)
-    mod/default-rules))
-(mod-fed-test!
-  "valid proof → not revoked"
-  (get mod-fed-rev-same :revoked)
-  false)
-
-;; exoneration invalidates the proof → revocation
-(define
-  mod-fed-exon-r
-  (mod/attach-evidence mod-fed-spam-r (mod/mk-evidence "exonerated" "mod")))
-(define
-  mod-fed-rev-inv
-  (mod/fed-revoke-if-invalidated
-    mod-fed-exon-r
-    mod-fed-spam-d
-    (list mod-fed-exon-r)
-    mod/default-rules))
-(mod-fed-test!
-  "invalidated proof → revoked"
-  (get mod-fed-rev-inv :revoked)
-  true)
-(mod-fed-test!
-  "re-decision after exoneration is keep"
-  (get (get mod-fed-rev-inv :decision) :action)
-  "keep")
-
-(define mod-fed-tests-run! (fn () {:failures mod-fed-failures :total mod-fed-count :passed mod-fed-pass :failed mod-fed-fail}))

lib/mod/tests/link.sx

@@ -1,86 +0,0 @@
-;; lib/mod/tests/link.sx — Ext 4: report linking + dedup.
-
-(define mod-lnk-count 0)
-(define mod-lnk-pass 0)
-(define mod-lnk-fail 0)
-(define mod-lnk-failures (list))
-
-(define
-  mod-lnk-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-lnk-count (+ mod-lnk-count 1))
-      (if
-        (= got expected)
-        (set! mod-lnk-pass (+ mod-lnk-pass 1))
-        (begin
-          (set! mod-lnk-fail (+ mod-lnk-fail 1))
-          (append!
-            mod-lnk-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── link-key + dedup ──
-
-(define mod-lnk-a (mod/mk-report "r1" "alice" "bob" "this is spam"))
-(define mod-lnk-a2 (mod/mk-report "r2" "alice" "bob" "THIS IS SPAM"))
-(define mod-lnk-b (mod/mk-report "r3" "carol" "bob" "abuse"))
-(define mod-lnk-c (mod/mk-report "r4" "alice" "eve" "this is spam"))
-
-(mod-lnk-test!
-  "identical reports share a link key (case-insensitive reason)"
-  (= (mod/link-key mod-lnk-a) (mod/link-key mod-lnk-a2))
-  true)
-(mod-lnk-test!
-  "different reporter → different key"
-  (= (mod/link-key mod-lnk-a) (mod/link-key mod-lnk-b))
-  false)
-(mod-lnk-test!
-  "different subject → different key"
-  (= (mod/link-key mod-lnk-a) (mod/link-key mod-lnk-c))
-  false)
-
-(define mod-lnk-set (list mod-lnk-a mod-lnk-a2 mod-lnk-b mod-lnk-c))
-(mod-lnk-test!
-  "dedup collapses identical reports"
-  (len (mod/dedup-reports mod-lnk-set))
-  3)
-(mod-lnk-test!
-  "duplicate-count counts collapsed"
-  (mod/duplicate-count mod-lnk-set)
-  1)
-(mod-lnk-test!
-  "dedup of all-distinct keeps all"
-  (len (mod/dedup-reports (list mod-lnk-a mod-lnk-b mod-lnk-c)))
-  3)
-
-;; ── Prolog-backed relational linking ──
-
-(mod-lnk-test!
-  "related-ids finds all reports about subject"
-  (len (mod/related-ids "bob" mod-lnk-set))
-  3)
-(mod-lnk-test!
-  "related-ids returns the ids"
-  (mod/related-ids "eve" mod-lnk-set)
-  (list "r4"))
-(mod-lnk-test!
-  "related-ids empty for unknown subject"
-  (mod/related-ids "nobody" mod-lnk-set)
-  (list))
-
-;; reporters: bob reported by alice (x2) + carol → 3 raw, 2 distinct
-(mod-lnk-test!
-  "reporters-of counts all reports"
-  (len (mod/reporters-of "bob" mod-lnk-set))
-  3)
-(mod-lnk-test!
-  "distinct reporters-of dedups reporters"
-  (len (mod/distinct-reporters-of "bob" mod-lnk-set))
-  2)
-(mod-lnk-test!
-  "distinct utility removes dups"
-  (mod/distinct (list "a" "b" "a" "c" "b"))
-  (list "a" "b" "c"))
-
-(define mod-link-tests-run! (fn () {:failures mod-lnk-failures :total mod-lnk-count :passed mod-lnk-pass :failed mod-lnk-fail}))

lib/mod/tests/lint.sx

@@ -1,122 +0,0 @@
-;; lib/mod/tests/lint.sx — Ext 5: policy rule-set static analysis.
-
-(define mod-lint-count 0)
-(define mod-lint-pass 0)
-(define mod-lint-fail 0)
-(define mod-lint-failures (list))
-
-(define
-  mod-lint-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-lint-count (+ mod-lint-count 1))
-      (if
-        (= got expected)
-        (set! mod-lint-pass (+ mod-lint-pass 1))
-        (begin
-          (set! mod-lint-fail (+ mod-lint-fail 1))
-          (append!
-            mod-lint-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── the default rule set is well-formed ──
-
-(mod-lint-test!
-  "default rules: no unreachable"
-  (mod/unreachable-rules mod/default-rules)
-  (list))
-(mod-lint-test!
-  "default rules: has catch-all"
-  (mod/has-catchall? mod/default-rules)
-  true)
-(mod-lint-test!
-  "default rules: no duplicate names"
-  (mod/duplicate-rule-names mod/default-rules)
-  (list))
-(mod-lint-test!
-  "default rules: well-formed"
-  (mod/rules-ok? mod/default-rules)
-  true)
-
-;; ── unreachable detection ──
-
-(define
-  mod-lint-shadowed
-  (list
-    (mod/mk-rule "spam-hide" :hide (list (list :classification "spam")))
-    (mod/mk-rule "catch-all" :keep (list))
-    (mod/mk-rule
-      "abuse-remove"
-      :remove (list (list :classification "abuse")))
-    (mod/mk-rule
-      "repeated"
-      :escalate (list (list :count-at-least 3)))))
-
-(mod-lint-test!
-  "rules after catch-all are unreachable"
-  (mod/unreachable-rules mod-lint-shadowed)
-  (list "abuse-remove" "repeated"))
-(mod-lint-test!
-  "shadowed rule set is not ok"
-  (mod/rules-ok? mod-lint-shadowed)
-  false)
-
-;; ── missing catch-all ──
-
-(define
-  mod-lint-nocatch
-  (list
-    (mod/mk-rule "spam-hide" :hide (list (list :classification "spam")))
-    (mod/mk-rule
-      "abuse-remove"
-      :remove (list (list :classification "abuse")))))
-
-(mod-lint-test!
-  "no catch-all detected"
-  (mod/has-catchall? mod-lint-nocatch)
-  false)
-(mod-lint-test!
-  "no unreachable when no catch-all"
-  (mod/unreachable-rules mod-lint-nocatch)
-  (list))
-(mod-lint-test!
-  "no-catch-all rule set is not ok"
-  (mod/rules-ok? mod-lint-nocatch)
-  false)
-
-;; ── duplicate names ──
-
-(define
-  mod-lint-dups
-  (list
-    (mod/mk-rule "x" :hide (list (list :classification "spam")))
-    (mod/mk-rule "x" :remove (list (list :classification "abuse")))
-    (mod/mk-rule "default" :keep (list))))
-
-(mod-lint-test!
-  "duplicate names detected"
-  (mod/duplicate-rule-names mod-lint-dups)
-  (list "x"))
-(mod-lint-test!
-  "duplicate-name rule set is not ok"
-  (mod/rules-ok? mod-lint-dups)
-  false)
-
-;; ── helpers ──
-
-(mod-lint-test!
-  "rule-unconditional? true for empty when"
-  (mod/rule-unconditional? (mod/mk-rule "d" :keep (list)))
-  true)
-(mod-lint-test!
-  "rule-unconditional? false with conditions"
-  (mod/rule-unconditional?
-    (mod/mk-rule "s" :hide (list (list :classification "spam"))))
-  false)
-(mod-lint-test!
-  "count-eq counts occurrences"
-  (mod/count-eq "a" (list "a" "b" "a"))
-  2)
-
-(define mod-lint-tests-run! (fn () {:failures mod-lint-failures :total mod-lint-count :passed mod-lint-pass :failed mod-lint-fail}))

lib/mod/tests/offenders.sx

@@ -1,115 +0,0 @@
-;; lib/mod/tests/offenders.sx — Ext 7: repeat-offender escalation.
-
-(define mod-off-count 0)
-(define mod-off-pass 0)
-(define mod-off-fail 0)
-(define mod-off-failures (list))
-
-(define
-  mod-off-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-off-count (+ mod-off-count 1))
-      (if
-        (= got expected)
-        (set! mod-off-pass (+ mod-off-pass 1))
-        (begin
-          (set! mod-off-fail (+ mod-off-fail 1))
-          (append!
-            mod-off-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── sanction? predicate ──
-
-(mod-off-test! "hide is a sanction" (mod/sanction? "hide") true)
-(mod-off-test! "remove is a sanction" (mod/sanction? "remove") true)
-(mod-off-test! "ban is a sanction" (mod/sanction? "ban") true)
-(mod-off-test! "keep is not a sanction" (mod/sanction? "keep") false)
-(mod-off-test! "escalate is not a sanction" (mod/sanction? "escalate") false)
-
-;; ── repeat-offender escalation over the audit log ──
-
-(mod/reset!)
-(mod/report "u1" "spammer" "this is spam")
-(mod/report "u2" "spammer" "buy now offer")
-(mod/report "u3" "spammer" "click here free money")
-(mod/report "u4" "innocent" "fine post")
-
-(mod-off-test!
-  "no sanctions before any decision"
-  (mod/subject-sanctions "spammer")
-  0)
-
-(define mod-off-d1 (mod/decide-escalating "r1" 2))
-(mod-off-test!
-  "first spam → hide (0 priors)"
-  (get mod-off-d1 :action)
-  "hide")
-(mod-off-test!
-  "one sanction recorded"
-  (mod/subject-sanctions "spammer")
-  1)
-
-(define mod-off-d2 (mod/decide-escalating "r2" 2))
-(mod-off-test!
-  "second spam → hide (1 prior, below k=2)"
-  (get mod-off-d2 :action)
-  "hide")
-(mod-off-test!
-  "two sanctions recorded"
-  (mod/subject-sanctions "spammer")
-  2)
-
-(define mod-off-d3 (mod/decide-escalating "r3" 2))
-(mod-off-test!
-  "third spam → ban (2 priors ≥ k)"
-  (get mod-off-d3 :action)
-  "ban")
-(mod-off-test!
-  "ban decision names repeat-offender rule"
-  (get mod-off-d3 :rule)
-  "repeat-offender-ban")
-(mod-off-test!
-  "ban proof records prior sanction count"
-  (get (get mod-off-d3 :proof) :prior-sanctions)
-  2)
-
-;; ── different subjects accumulate independently ──
-
-(define mod-off-d4 (mod/decide-escalating "r4" 2))
-(mod-off-test!
-  "innocent keep → not escalated"
-  (get mod-off-d4 :action)
-  "keep")
-(mod-off-test!
-  "innocent has no sanctions"
-  (mod/subject-sanctions "innocent")
-  0)
-(mod-off-test!
-  "repeat-offender? true for spammer at k=2"
-  (mod/repeat-offender? "spammer" 2)
-  true)
-(mod-off-test!
-  "repeat-offender? false for innocent at k=1"
-  (mod/repeat-offender? "innocent" 1)
-  false)
-
-;; ── non-sanction decisions are never upgraded to ban ──
-;; r5 is a clean post, but it is the 4th report about "spammer", so the
-;; repeated-report rule escalates it. escalate is not a sanction, so it passes
-;; through decide-escalating unchanged (never becomes :ban).
-
-(mod/report "u5" "spammer" "a perfectly fine post")
-(define mod-off-d5 (mod/decide-escalating "r5" 1))
-(mod-off-test!
-  "non-sanction (escalate) decision is not upgraded to ban"
-  (get mod-off-d5 :action)
-  "escalate")
-
-(mod-off-test!
-  "decide-escalating unknown id → nil"
-  (mod/decide-escalating "r99" 2)
-  nil)
-
-(define mod-offenders-tests-run! (fn () {:failures mod-off-failures :total mod-off-count :passed mod-off-pass :failed mod-off-fail}))

lib/mod/tests/pipeline.sx

@@ -1,112 +0,0 @@
-;; lib/mod/tests/pipeline.sx — Ext 19: end-to-end triage orchestration.
-
-(define mod-pp-count 0)
-(define mod-pp-pass 0)
-(define mod-pp-fail 0)
-(define mod-pp-failures (list))
-
-(define
-  mod-pp-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-pp-count (+ mod-pp-count 1))
-      (if
-        (= got expected)
-        (set! mod-pp-pass (+ mod-pp-pass 1))
-        (begin
-          (set! mod-pp-fail (+ mod-pp-fail 1))
-          (append!
-            mod-pp-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(mod/policies-reset!)
-(mod/register-policy!
-  "market"
-  (mod/ruleset
-    (mod/defrule "market-spam-remove" :remove (list :classification "spam"))
-    (mod/defrule "default-keep" :keep)))
-
-;; ── spam in the market domain: full bundle ──
-
-(define mod-pp-spam (mod/mk-report "r1" "u" "bob" "this is spam"))
-(define
-  mod-pp
-  (mod/triage-pipeline "market" mod-pp-spam (list mod-pp-spam) "inst.example"))
-
-(mod-pp-test!
-  "pipeline action (market policy → remove)"
-  (mod/pipeline-action mod-pp)
-  "remove")
-(mod-pp-test! "pipeline rule" (get mod-pp :rule) "market-spam-remove")
-(mod-pp-test!
-  "pipeline explanation mentions the action"
-  (mod/str-contains? (get mod-pp :explanation) "remove")
-  true)
-(mod-pp-test!
-  "pipeline activity is Delete (remove)"
-  (get (mod/pipeline-activity mod-pp) :type)
-  "Delete")
-(mod-pp-test!
-  "pipeline activity object is the report"
-  (get (mod/pipeline-activity mod-pp) :object)
-  "r1")
-(mod-pp-test!
-  "pipeline wire round-trips to the same action"
-  (get (mod/wire->decision (mod/pipeline-wire mod-pp)) :action)
-  "remove")
-
-;; ── same report, blog domain (default) → hide, Flag ──
-
-(define
-  mod-pp-blog
-  (mod/triage-pipeline "blog" mod-pp-spam (list mod-pp-spam) "inst.example"))
-(mod-pp-test!
-  "blog default policy → hide"
-  (mod/pipeline-action mod-pp-blog)
-  "hide")
-(mod-pp-test!
-  "blog activity is Flag"
-  (get (mod/pipeline-activity mod-pp-blog) :type)
-  "Flag")
-
-;; ── clean report: keep, no activity, explanation says (none) ──
-
-(define mod-pp-clean (mod/mk-report "r2" "u" "eve" "a fine post"))
-(define
-  mod-pp-k
-  (mod/triage-pipeline
-    "market"
-    mod-pp-clean
-    (list mod-pp-clean)
-    "inst.example"))
-(mod-pp-test! "clean → keep" (mod/pipeline-action mod-pp-k) "keep")
-(mod-pp-test! "keep → no activity" (mod/pipeline-activity mod-pp-k) nil)
-(mod-pp-test!
-  "keep explanation says no evidence"
-  (mod/str-contains? (get mod-pp-k :explanation) "Evidence: (none)")
-  true)
-(mod-pp-test!
-  "keep wire still round-trips"
-  (get (mod/wire->decision (mod/pipeline-wire mod-pp-k)) :rule)
-  "default-keep")
-
-;; ── federated handoff: market decision crosses to a peer, trust-gated ──
-
-(mod/fed-reset!)
-(define mod-pp-peer-dec (mod/wire->decision (mod/pipeline-wire mod-pp)))
-(mod-pp-test!
-  "untrusted peer: market decision is advisory"
-  (get (mod/fed-receive-decision "peerX" mod-pp-peer-dec) :applied)
-  false)
-(mod/grant-trust "peerY" :mod)
-(mod-pp-test!
-  "trusted peer: market decision applies"
-  (get (mod/fed-receive-decision "peerY" mod-pp-peer-dec) :applied)
-  true)
-(mod-pp-test!
-  "applied action is remove"
-  (get (mod/fed-applied-action "r1") :action)
-  "remove")
-
-(define mod-pipeline-tests-run! (fn () {:failures mod-pp-failures :total mod-pp-count :passed mod-pp-pass :failed mod-pp-fail}))

lib/mod/tests/policies.sx

@@ -1,112 +0,0 @@
-;; lib/mod/tests/policies.sx — Ext 17: per-domain policy registry.
-
-(define mod-pol-count 0)
-(define mod-pol-pass 0)
-(define mod-pol-fail 0)
-(define mod-pol-failures (list))
-
-(define
-  mod-pol-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-pol-count (+ mod-pol-count 1))
-      (if
-        (= got expected)
-        (set! mod-pol-pass (+ mod-pol-pass 1))
-        (begin
-          (set! mod-pol-fail (+ mod-pol-fail 1))
-          (append!
-            mod-pol-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(mod/policies-reset!)
-
-;; market is strict: spam is removed outright, not just hidden
-(define
-  mod-pol-market-rules
-  (list
-    (mod/mk-rule
-      "market-spam-remove"
-      :remove (list (list :classification "spam")))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(mod-pol-test!
-  "unregistered domain falls back to default"
-  (mod/policy-registered? "market")
-  false)
-(mod/register-policy! "market" mod-pol-market-rules)
-(mod-pol-test!
-  "domain registered after register!"
-  (mod/policy-registered? "market")
-  true)
-
-(define mod-pol-spam (mod/mk-report "r1" "a" "b" "this is spam"))
-
-;; ── same report, different domain → different action ──
-
-(mod-pol-test!
-  "market policy removes spam"
-  (get (mod/decide-in "market" mod-pol-spam (list mod-pol-spam)) :action)
-  "remove")
-(mod-pol-test!
-  "market decision uses market rule"
-  (get (mod/decide-in "market" mod-pol-spam (list mod-pol-spam)) :rule)
-  "market-spam-remove")
-(mod-pol-test!
-  "blog (unregistered) uses default → hide"
-  (get (mod/decide-in "blog" mod-pol-spam (list mod-pol-spam)) :action)
-  "hide")
-(mod-pol-test!
-  "blog decision uses default rule"
-  (get (mod/decide-in "blog" mod-pol-spam (list mod-pol-spam)) :rule)
-  "spam-hide")
-
-;; ── policy-for resolution ──
-
-(mod-pol-test!
-  "policy-for market returns market rules"
-  (mod/policy-for "market")
-  mod-pol-market-rules)
-(mod-pol-test!
-  "policy-for unknown returns default"
-  (mod/policy-for "events")
-  mod/default-rules)
-(mod-pol-test!
-  "registered-domains lists market"
-  (mod/registered-domains)
-  (list "market"))
-
-;; ── a second domain ──
-
-(define
-  mod-pol-events-rules
-  (list (mod/mk-rule "events-keep-all" :keep (list))))
-
-(mod/register-policy! "events" mod-pol-events-rules)
-(mod-pol-test!
-  "events policy keeps everything (even spam)"
-  (get (mod/decide-in "events" mod-pol-spam (list mod-pol-spam)) :action)
-  "keep")
-(mod-pol-test!
-  "two domains registered"
-  (len (mod/registered-domains))
-  2)
-(mod-pol-test!
-  "market still removes after second registration"
-  (get (mod/decide-in "market" mod-pol-spam (list mod-pol-spam)) :action)
-  "remove")
-
-;; ── clean report is keep everywhere ──
-
-(define mod-pol-clean (mod/mk-report "r2" "a" "b" "a fine post"))
-(mod-pol-test!
-  "clean report keep in market"
-  (get (mod/decide-in "market" mod-pol-clean (list mod-pol-clean)) :action)
-  "keep")
-(mod-pol-test!
-  "clean report keep in blog"
-  (get (mod/decide-in "blog" mod-pol-clean (list mod-pol-clean)) :action)
-  "keep")
-
-(define mod-policies-tests-run! (fn () {:failures mod-pol-failures :total mod-pol-count :passed mod-pol-pass :failed mod-pol-fail}))

lib/mod/tests/quorum.sx

@@ -1,119 +0,0 @@
-;; lib/mod/tests/quorum.sx — Ext 8: quorum over distinct reporters.
-
-(define mod-q-count 0)
-(define mod-q-pass 0)
-(define mod-q-fail 0)
-(define mod-q-failures (list))
-
-(define
-  mod-q-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-q-count (+ mod-q-count 1))
-      (if
-        (= got expected)
-        (set! mod-q-pass (+ mod-q-pass 1))
-        (begin
-          (set! mod-q-fail (+ mod-q-fail 1))
-          (append!
-            mod-q-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(define
-  mod-q-rules
-  (list
-    (mod/mk-rule
-      "quorum-hide"
-      :hide (list (list :reporters-at-least 2)))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-;; ── two distinct reporters meet quorum ──
-
-(define
-  mod-q-two
-  (list
-    (mod/mk-report "r1" "alice" "bob" "off-topic")
-    (mod/mk-report "r2" "carol" "bob" "off-topic")))
-
-(mod-q-test!
-  "two distinct reporters → hide"
-  (get (mod/decide-quorum (first mod-q-two) mod-q-two mod-q-rules) :action)
-  "hide")
-(mod-q-test!
-  "quorum decision names the rule"
-  (get (mod/decide-quorum (first mod-q-two) mod-q-two mod-q-rules) :rule)
-  "quorum-hide")
-(mod-q-test!
-  "quorum decision tagged strategy"
-  (get (mod/decide-quorum (first mod-q-two) mod-q-two mod-q-rules) :strategy)
-  "quorum")
-
-;; ── single reporter does not meet quorum ──
-
-(define mod-q-one (list (mod/mk-report "r1" "alice" "bob" "off-topic")))
-(mod-q-test!
-  "one reporter → keep (below quorum)"
-  (get (mod/decide-quorum (first mod-q-one) mod-q-one mod-q-rules) :action)
-  "keep")
-
-;; ── anti-brigade: one user filing many reports does NOT meet quorum ──
-
-(define
-  mod-q-brigade
-  (list
-    (mod/mk-report "r1" "alice" "bob" "off-topic")
-    (mod/mk-report "r2" "alice" "bob" "off-topic")
-    (mod/mk-report "r3" "alice" "bob" "off-topic")))
-
-(mod-q-test!
-  "three reports, one reporter → keep (quorum counts distinct)"
-  (get
-    (mod/decide-quorum (first mod-q-brigade) mod-q-brigade mod-q-rules)
-    :action)
-  "keep")
-
-;; contrast: the count rule WOULD fire on the same brigade (3 reports ≥ 3) —
-;; quorum is strictly stronger against single-actor brigading
-(mod-q-test!
-  "count rule fires on the brigade (distinct from quorum)"
-  (get
-    (mod/decide-report (first mod-q-brigade) mod-q-brigade mod/default-rules)
-    :action)
-  "escalate")
-
-;; ── three distinct reporters ──
-
-(define
-  mod-q-three
-  (list
-    (mod/mk-report "r1" "alice" "bob" "off-topic")
-    (mod/mk-report "r2" "carol" "bob" "off-topic")
-    (mod/mk-report "r3" "dave" "bob" "off-topic")))
-
-(mod-q-test!
-  "three distinct reporters → hide"
-  (get
-    (mod/decide-quorum (first mod-q-three) mod-q-three mod-q-rules)
-    :action)
-  "hide")
-(mod-q-test!
-  "quorum proof goal solved"
-  (get
-    (first
-      (get
-        (get
-          (mod/decide-quorum (first mod-q-three) mod-q-three mod-q-rules)
-          :proof)
-        :goals))
-    :solved)
-  true)
-
-;; ── cond->goal compiles :reporters-at-least ──
-
-(mod-q-test!
-  "cond->goal :reporters-at-least"
-  (mod/cond->goal (list :reporters-at-least 2) "Id")
-  "report(Id, _, Sr), setof(Br, report(_, Br, Sr), Bsr), length(Bsr, Nr), Nr >= 2")
-
-(define mod-quorum-tests-run! (fn () {:failures mod-q-failures :total mod-q-count :passed mod-q-pass :failed mod-q-fail}))

lib/mod/tests/severity.sx

@@ -1,120 +0,0 @@
-;; lib/mod/tests/severity.sx — Ext 6: strictest-wins decision strategy.
-
-(define mod-sev-count 0)
-(define mod-sev-pass 0)
-(define mod-sev-fail 0)
-(define mod-sev-failures (list))
-
-(define
-  mod-sev-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-sev-count (+ mod-sev-count 1))
-      (if
-        (= got expected)
-        (set! mod-sev-pass (+ mod-sev-pass 1))
-        (begin
-          (set! mod-sev-fail (+ mod-sev-fail 1))
-          (append!
-            mod-sev-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── severity ranking ──
-
-(mod-sev-test! "ban most severe" (mod/action-severity "ban") 4)
-(mod-sev-test!
-  "remove > hide"
-  (< (mod/action-severity "hide") (mod/action-severity "remove"))
-  true)
-(mod-sev-test! "keep least severe" (mod/action-severity "keep") 0)
-(mod-sev-test!
-  "escalate above keep"
-  (< (mod/action-severity "keep") (mod/action-severity "escalate"))
-  true)
-
-;; ── strictest agrees with default-rules on simple cases ──
-
-(define mod-sev-spam (mod/mk-report "r1" "a" "b" "this is spam"))
-(mod-sev-test!
-  "strictest spam → hide"
-  (get
-    (mod/decide-strictest mod-sev-spam (list mod-sev-spam) mod/default-rules)
-    :action)
-  "hide")
-(define mod-sev-clean (mod/mk-report "r2" "a" "b" "a fine post"))
-(mod-sev-test!
-  "strictest clean → keep"
-  (get
-    (mod/decide-strictest
-      mod-sev-clean
-      (list mod-sev-clean)
-      mod/default-rules)
-    :action)
-  "keep")
-(mod-sev-test!
-  "decision tagged strategy strictest"
-  (get
-    (mod/decide-strictest mod-sev-spam (list mod-sev-spam) mod/default-rules)
-    :strategy)
-  "strictest")
-
-;; ── strictest diverges from first-match when order ≠ severity ──
-
-(define
-  mod-sev-rules
-  (list
-    (mod/mk-rule
-      "early-escalate"
-      :escalate (list (list :count-at-least 1)))
-    (mod/mk-rule "spam-remove" :remove (list (list :classification "spam")))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define mod-sev-r (mod/mk-report "r3" "a" "b" "this is spam"))
-
-(mod-sev-test!
-  "first-match picks earliest rule (escalate)"
-  (get (mod/decide-report mod-sev-r (list mod-sev-r) mod-sev-rules) :action)
-  "escalate")
-(mod-sev-test!
-  "strictest picks harshest action (remove)"
-  (get
-    (mod/decide-strictest mod-sev-r (list mod-sev-r) mod-sev-rules)
-    :action)
-  "remove")
-(mod-sev-test!
-  "strictest names the harshest rule"
-  (get (mod/decide-strictest mod-sev-r (list mod-sev-r) mod-sev-rules) :rule)
-  "spam-remove")
-(mod-sev-test!
-  "strictest carries proof goals"
-  (len
-    (get
-      (get
-        (mod/decide-strictest mod-sev-r (list mod-sev-r) mod-sev-rules)
-        :proof)
-      :goals))
-  1)
-
-;; ── strictest among three matches (spam + repeated) ──
-
-(define mod-sev-rep (mod/mk-report "r4" "a" "b" "buy now spam"))
-(define mod-sev-reps (list mod-sev-rep mod-sev-rep mod-sev-rep))
-(mod-sev-test!
-  "strictest among hide+escalate+keep → hide (default rules)"
-  (get
-    (mod/decide-strictest mod-sev-rep mod-sev-reps mod/default-rules)
-    :action)
-  "hide")
-
-;; ── strictest-sol helper ──
-
-(mod-sev-test!
-  "strictest-sol picks max severity"
-  (dict-get
-    (mod/strictest-sol (list {:Action "keep" :Rule "k"} {:Action "remove" :Rule "r"} {:Action "hide" :Rule "h"}))
-    "Action")
-  "remove")
-(mod-sev-test! "strictest-sol nil for empty" (mod/strictest-sol (list)) nil)
-
-(define mod-severity-tests-run! (fn () {:failures mod-sev-failures :total mod-sev-count :passed mod-sev-pass :failed mod-sev-fail}))

lib/mod/tests/sla.sx

@@ -1,108 +0,0 @@
-;; lib/mod/tests/sla.sx — Ext 13: SLA sweep over pending lifecycle cases.
-
-(define mod-sla-count 0)
-(define mod-sla-pass 0)
-(define mod-sla-fail 0)
-(define mod-sla-failures (list))
-
-(define
-  mod-sla-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-sla-count (+ mod-sla-count 1))
-      (if
-        (= got expected)
-        (set! mod-sla-pass (+ mod-sla-pass 1))
-        (begin
-          (set! mod-sla-fail (+ mod-sla-fail 1))
-          (append!
-            mod-sla-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── pending-state? ──
-
-(mod-sla-test! "open is pending" (mod/pending-state? "open") true)
-(mod-sla-test! "triaged is pending" (mod/pending-state? "triaged") true)
-(mod-sla-test! "appealed is pending" (mod/pending-state? "appealed") true)
-(mod-sla-test! "decided is not pending" (mod/pending-state? "decided") false)
-(mod-sla-test! "final is not pending" (mod/pending-state? "final") false)
-
-;; build cases in known states
-(define mod-sla-spam (mod/mk-report "r1" "u" "bob" "this is spam"))
-(define mod-sla-spam-reports (list mod-sla-spam))
-(define
-  mod-sla-triaged
-  (mod/case-triage
-    (mod/mk-case mod-sla-spam)
-    mod-sla-spam-reports
-    mod/default-rules))
-(define mod-sla-decided (mod/case-resolve mod-sla-triaged))
-(define mod-sla-open (mod/mk-case (mod/mk-report "r2" "u" "eve" "hello")))
-
-;; ── overdue? ──
-
-(define mod-sla-tc-old (mod/mk-timed-case mod-sla-triaged 0))
-(define mod-sla-tc-fresh (mod/mk-timed-case mod-sla-triaged 90))
-(define mod-sla-tc-done (mod/mk-timed-case mod-sla-decided 0))
-
-(mod-sla-test!
-  "old triaged case is overdue"
-  (mod/overdue? mod-sla-tc-old 100 50)
-  true)
-(mod-sla-test!
-  "fresh triaged case not overdue"
-  (mod/overdue? mod-sla-tc-fresh 100 50)
-  false)
-(mod-sla-test!
-  "decided case never overdue"
-  (mod/overdue? mod-sla-tc-done 100 50)
-  false)
-(mod-sla-test!
-  "age computes elapsed ticks"
-  (mod/age mod-sla-tc-old 100)
-  100)
-(mod-sla-test!
-  "boundary: exactly at deadline not overdue"
-  (mod/overdue?
-    (mod/mk-timed-case mod-sla-triaged 50)
-    100
-    50)
-  false)
-(mod-sla-test!
-  "boundary: one past deadline overdue"
-  (mod/overdue?
-    (mod/mk-timed-case mod-sla-triaged 49)
-    100
-    50)
-  true)
-
-;; ── sweep over a mixed queue ──
-
-(define
-  mod-sla-queue
-  (list
-    (mod/mk-timed-case mod-sla-triaged 0)
-    (mod/mk-timed-case mod-sla-decided 0)
-    (mod/mk-timed-case mod-sla-open 90)))      ;; r2, pending, age 10 → not
-
-(mod-sla-test!
-  "sweep finds only the overdue pending case"
-  (mod/sla-sweep mod-sla-queue 100 50)
-  (list "r1"))
-(mod-sla-test!
-  "overdue-count agrees"
-  (mod/overdue-count mod-sla-queue 100 50)
-  1)
-
-;; tighten deadline so the young open case also breaches
-(mod-sla-test!
-  "tighter deadline catches the open case too"
-  (mod/overdue-count mod-sla-queue 100 5)
-  2)
-(mod-sla-test!
-  "empty queue → no breaches"
-  (mod/sla-sweep (list) 100 50)
-  (list))
-
-(define mod-sla-tests-run! (fn () {:failures mod-sla-failures :total mod-sla-count :passed mod-sla-pass :failed mod-sla-fail}))

lib/mod/tests/temporal.sx

@@ -1,156 +0,0 @@
-;; lib/mod/tests/temporal.sx — Ext 12: burst detection over a time window.
-
-(define mod-tm-count 0)
-(define mod-tm-pass 0)
-(define mod-tm-fail 0)
-(define mod-tm-failures (list))
-
-(define
-  mod-tm-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-tm-count (+ mod-tm-count 1))
-      (if
-        (= got expected)
-        (set! mod-tm-pass (+ mod-tm-pass 1))
-        (begin
-          (set! mod-tm-fail (+ mod-tm-fail 1))
-          (append!
-            mod-tm-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(define
-  mod-tm-at
-  (fn (id about t) (mod/with-at (mod/mk-report id "u" about "off-topic") t)))
-
-(define
-  mod-tm-rules
-  (list
-    (mod/mk-rule "burst-hide" :hide (list (list :burst-at-least 3)))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-;; ── window-count helper ──
-
-(define
-  mod-tm-burst
-  (list
-    (mod-tm-at "r1" "bob" 10)
-    (mod-tm-at "r2" "bob" 11)
-    (mod-tm-at "r3" "bob" 12)))
-(define
-  mod-tm-slow
-  (list
-    (mod-tm-at "r1" "bob" 1)
-    (mod-tm-at "r2" "bob" 2)
-    (mod-tm-at "r3" "bob" 12)))
-
-(mod-tm-test!
-  "window-count: all 3 within window"
-  (mod/window-count "bob" mod-tm-burst 12 5)
-  3)
-(mod-tm-test!
-  "window-count: only 1 within window"
-  (mod/window-count "bob" mod-tm-slow 12 5)
-  1)
-(mod-tm-test!
-  "window-count: subject filter"
-  (mod/window-count "eve" mod-tm-burst 12 5)
-  0)
-
-;; ── burst fires; slow accumulation does not ──
-
-(mod-tm-test!
-  "burst (3 in window) → hide"
-  (get
-    (mod/decide-temporal
-      (first mod-tm-burst)
-      mod-tm-burst
-      mod-tm-rules
-      12
-      5)
-    :action)
-  "hide")
-(mod-tm-test!
-  "slow accumulation (1 in window) → keep"
-  (get
-    (mod/decide-temporal
-      (first mod-tm-slow)
-      mod-tm-slow
-      mod-tm-rules
-      12
-      5)
-    :action)
-  "keep")
-
-;; ── contrast: the plain count rule fires on BOTH (3 total reports) ──
-(mod-tm-test!
-  "count rule fires on slow case (distinct from burst)"
-  (get
-    (mod/decide-report (first mod-tm-slow) mod-tm-slow mod/default-rules)
-    :action)
-  "escalate")
-
-;; ── decision shape ──
-
-(define
-  mod-tm-d
-  (mod/decide-temporal
-    (first mod-tm-burst)
-    mod-tm-burst
-    mod-tm-rules
-    12
-    5))
-(mod-tm-test! "burst decision rule" (get mod-tm-d :rule) "burst-hide")
-(mod-tm-test!
-  "burst decision tagged strategy"
-  (get mod-tm-d :strategy)
-  "temporal")
-(mod-tm-test!
-  "burst recorded in proof"
-  (get (get mod-tm-d :proof) :burst)
-  3)
-(mod-tm-test!
-  "burst proof goal solved"
-  (get (first (get (get mod-tm-d :proof) :goals)) :solved)
-  true)
-
-;; ── window boundary is inclusive ──
-
-(define
-  mod-tm-edge
-  (list
-    (mod-tm-at "r1" "bob" 7)
-    (mod-tm-at "r2" "bob" 8)
-    (mod-tm-at "r3" "bob" 9)))
-(mod-tm-test!
-  "window boundary inclusive (now-window = at)"
-  (mod/window-count "bob" mod-tm-edge 12 5)
-  3)
-
-;; ── schema :at round-trips and survives evidence attach ──
-
-(mod-tm-test!
-  "report-at reads timestamp"
-  (mod/report-at (mod-tm-at "r1" "bob" 42))
-  42)
-(mod-tm-test!
-  "default report-at is 0"
-  (mod/report-at (mod/mk-report "r1" "a" "b" "x"))
-  0)
-(mod-tm-test!
-  "attach-evidence preserves :at"
-  (mod/report-at
-    (mod/attach-evidence
-      (mod-tm-at "r1" "bob" 42)
-      (mod/mk-evidence "k" "v")))
-  42)
-
-;; ── cond->goal :burst-at-least ──
-
-(mod-tm-test!
-  "cond->goal :burst-at-least"
-  (mod/cond->goal (list :burst-at-least 3) "Id")
-  "report(Id, _, Sb), burst_count(Sb, Nb), Nb >= 3")
-
-(define mod-temporal-tests-run! (fn () {:failures mod-tm-failures :total mod-tm-count :passed mod-tm-pass :failed mod-tm-fail}))

lib/mod/tests/trace.sx

@@ -1,116 +0,0 @@
-;; lib/mod/tests/trace.sx — Ext 9: policy dry-run diagnostics.
-
-(define mod-tr-count 0)
-(define mod-tr-pass 0)
-(define mod-tr-fail 0)
-(define mod-tr-failures (list))
-
-(define
-  mod-tr-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-tr-count (+ mod-tr-count 1))
-      (if
-        (= got expected)
-        (set! mod-tr-pass (+ mod-tr-pass 1))
-        (begin
-          (set! mod-tr-fail (+ mod-tr-fail 1))
-          (append!
-            mod-tr-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-(define
-  mod-tr-find
-  (fn
-    (trace nm)
-    (reduce (fn (acc t) (if (= (get t :rule) nm) t acc)) nil trace)))
-
-;; ── trace a spam report against the default rules ──
-
-(define mod-tr-spam (mod/mk-report "r1" "alice" "bob" "this is spam"))
-(define
-  mod-tr-t
-  (mod/trace-rules mod-tr-spam (list mod-tr-spam) mod/default-rules))
-
-(mod-tr-test! "trace covers every rule" (len mod-tr-t) 6)
-(mod-tr-test!
-  "spam-hide fires"
-  (get (mod-tr-find mod-tr-t "spam-hide") :proved)
-  true)
-(mod-tr-test!
-  "default-keep always fires"
-  (get (mod-tr-find mod-tr-t "default-keep") :proved)
-  true)
-(mod-tr-test!
-  "reviewer-remove does not fire (no evidence)"
-  (get (mod-tr-find mod-tr-t "reviewer-remove") :proved)
-  false)
-(mod-tr-test!
-  "exonerated-keep does not fire"
-  (get (mod-tr-find mod-tr-t "exonerated-keep") :proved)
-  false)
-(mod-tr-test!
-  "abuse-remove does not fire"
-  (get (mod-tr-find mod-tr-t "abuse-remove") :proved)
-  false)
-
-;; ── winner matches the engine ──
-
-(mod-tr-test!
-  "first-proved is spam-hide"
-  (get (mod/first-proved mod-tr-t) :rule)
-  "spam-hide")
-(mod-tr-test!
-  "winner action matches decide-report"
-  (get (mod/first-proved mod-tr-t) :action)
-  (get
-    (mod/decide-report mod-tr-spam (list mod-tr-spam) mod/default-rules)
-    :action))
-
-;; ── an unproved rule shows which goal failed ──
-
-(define
-  mod-tr-rev-goals
-  (get (mod-tr-find mod-tr-t "reviewer-remove") :goals))
-(mod-tr-test!
-  "reviewer-remove goal is unsolved"
-  (get (first mod-tr-rev-goals) :solved)
-  false)
-(define mod-tr-spam-goals (get (mod-tr-find mod-tr-t "spam-hide") :goals))
-(mod-tr-test!
-  "spam-hide goal is solved"
-  (get (first mod-tr-spam-goals) :solved)
-  true)
-
-;; ── proved-rules list + rendering ──
-
-(mod-tr-test!
-  "proved-rules lists fired rules in order"
-  (mod/proved-rules mod-tr-t)
-  (list "spam-hide" "default-keep"))
-(mod-tr-test!
-  "trace-report marks a firing rule"
-  (mod/str-contains? (mod/trace-report mod-tr-t) "[fires] spam-hide")
-  true)
-(mod-tr-test!
-  "trace-report marks a non-firing rule"
-  (mod/str-contains? (mod/trace-report mod-tr-t) "[  -  ] reviewer-remove")
-  true)
-
-;; ── clean report: only default-keep fires ──
-
-(define mod-tr-clean (mod/mk-report "r2" "a" "b" "a fine post"))
-(define
-  mod-tr-tc
-  (mod/trace-rules mod-tr-clean (list mod-tr-clean) mod/default-rules))
-(mod-tr-test!
-  "clean report: only default-keep proves"
-  (mod/proved-rules mod-tr-tc)
-  (list "default-keep"))
-(mod-tr-test!
-  "clean report winner is default-keep"
-  (get (mod/first-proved mod-tr-tc) :rule)
-  "default-keep")
-
-(define mod-trace-tests-run! (fn () {:failures mod-tr-failures :total mod-tr-count :passed mod-tr-pass :failed mod-tr-fail}))

lib/mod/tests/whatif.sx

@@ -1,117 +0,0 @@
-;; lib/mod/tests/whatif.sx — Ext 10: policy what-if / impact analysis.
-
-(define mod-wi-count 0)
-(define mod-wi-pass 0)
-(define mod-wi-fail 0)
-(define mod-wi-failures (list))
-
-(define
-  mod-wi-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-wi-count (+ mod-wi-count 1))
-      (if
-        (= got expected)
-        (set! mod-wi-pass (+ mod-wi-pass 1))
-        (begin
-          (set! mod-wi-fail (+ mod-wi-fail 1))
-          (append!
-            mod-wi-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; rules-b is the default policy with spam-hide removed: spam now falls through
-;; to default-keep. A spam report flips hide → keep; everything else is unchanged.
-(define mod-wi-rules-a mod/default-rules)
-(define
-  mod-wi-rules-b
-  (list
-    (mod/mk-rule
-      "reviewer-remove"
-      :remove (list (list :evidence "confirmed-abuse")))
-    (mod/mk-rule
-      "abuse-remove"
-      :remove (list (list :classification "abuse")))
-    (mod/mk-rule
-      "repeated-escalate"
-      :escalate (list (list :count-at-least 3)))
-    (mod/mk-rule "default-keep" :keep (list))))
-
-(define mod-wi-spam (mod/mk-report "r1" "a" "bob" "this is spam"))
-(define mod-wi-abuse (mod/mk-report "r2" "a" "carol" "harassment here"))
-(define mod-wi-clean (mod/mk-report "r3" "a" "dave" "a fine post"))
-
-;; ── single-report diff ──
-
-(define
-  mod-wi-d
-  (mod/decision-diff
-    mod-wi-spam
-    (list mod-wi-spam)
-    mod-wi-rules-a
-    mod-wi-rules-b))
-(mod-wi-test! "spam before = hide" (get mod-wi-d :before) "hide")
-(mod-wi-test! "spam after = keep" (get mod-wi-d :after) "keep")
-(mod-wi-test! "spam decision flips" (get mod-wi-d :changed) true)
-(mod-wi-test! "diff carries report id" (get mod-wi-d :report-id) "r1")
-
-(define
-  mod-wi-da
-  (mod/decision-diff
-    mod-wi-abuse
-    (list mod-wi-abuse)
-    mod-wi-rules-a
-    mod-wi-rules-b))
-(mod-wi-test! "abuse unchanged (remove both)" (get mod-wi-da :changed) false)
-(mod-wi-test! "abuse stays remove" (get mod-wi-da :after) "remove")
-
-(define
-  mod-wi-dc
-  (mod/decision-diff
-    mod-wi-clean
-    (list mod-wi-clean)
-    mod-wi-rules-a
-    mod-wi-rules-b))
-(mod-wi-test! "clean unchanged (keep both)" (get mod-wi-dc :changed) false)
-
-;; ── batch impact ──
-
-(define mod-wi-batch (list mod-wi-spam mod-wi-abuse mod-wi-clean))
-(define
-  mod-wi-impact
-  (mod/policy-impact mod-wi-batch mod-wi-rules-a mod-wi-rules-b))
-
-(mod-wi-test!
-  "impact lists only changed reports"
-  (len mod-wi-impact)
-  1)
-(mod-wi-test!
-  "impacted report is the spam one"
-  (get (first mod-wi-impact) :report-id)
-  "r1")
-(mod-wi-test!
-  "impact-count agrees"
-  (mod/impact-count mod-wi-batch mod-wi-rules-a mod-wi-rules-b)
-  1)
-
-;; ── identical rule sets → no impact ──
-
-(mod-wi-test!
-  "same rules → zero impact"
-  (mod/impact-count mod-wi-batch mod-wi-rules-a mod-wi-rules-a)
-  0)
-(mod-wi-test!
-  "same rules → empty report"
-  (mod/impact-report mod-wi-batch mod-wi-rules-a mod-wi-rules-a)
-  "No decisions change.")
-
-;; ── rendering ──
-
-(mod-wi-test!
-  "impact-report renders the flip"
-  (mod/str-contains?
-    (mod/impact-report mod-wi-batch mod-wi-rules-a mod-wi-rules-b)
-    "r1: hide → keep")
-  true)
-
-(define mod-whatif-tests-run! (fn () {:failures mod-wi-failures :total mod-wi-count :passed mod-wi-pass :failed mod-wi-fail}))

lib/mod/tests/wire.sx

@@ -1,96 +0,0 @@
-;; lib/mod/tests/wire.sx — Ext 14: decision wire format + federated transport.
-
-(define mod-w-count 0)
-(define mod-w-pass 0)
-(define mod-w-fail 0)
-(define mod-w-failures (list))
-
-(define
-  mod-w-test!
-  (fn
-    (name got expected)
-    (begin
-      (set! mod-w-count (+ mod-w-count 1))
-      (if
-        (= got expected)
-        (set! mod-w-pass (+ mod-w-pass 1))
-        (begin
-          (set! mod-w-fail (+ mod-w-fail 1))
-          (append!
-            mod-w-failures
-            (str name "\n    expected: " expected "\n    got:      " got)))))))
-
-;; ── split-char ──
-
-(mod-w-test! "split on pipe" (mod/split-char "a|b|c" "|") (list "a" "b" "c"))
-(mod-w-test! "split single field" (mod/split-char "abc" "|") (list "abc"))
-(mod-w-test!
-  "split four fields"
-  (len (mod/split-char "MOD1|r1|hide|spam-hide" "|"))
-  4)
-
-;; ── serialize ──
-
-(define
-  mod-w-dec
-  (mod/decide-report
-    (mod/mk-report "r1" "a" "bob" "this is spam")
-    (list (mod/mk-report "r1" "a" "bob" "this is spam"))
-    mod/default-rules))
-(define mod-w-line (mod/decision->wire mod-w-dec))
-
-(mod-w-test!
-  "wire is versioned + delimited"
-  mod-w-line
-  "MOD1|r1|hide|spam-hide")
-(mod-w-test!
-  "wire-valid? accepts well-formed"
-  (mod/wire-valid? mod-w-line)
-  true)
-(mod-w-test!
-  "wire-valid? rejects junk"
-  (mod/wire-valid? "not a wire line")
-  false)
-(mod-w-test!
-  "wire-valid? rejects wrong version"
-  (mod/wire-valid? "MOD9|r1|hide|x")
-  false)
-
-;; ── round-trip ──
-
-(define mod-w-back (mod/wire->decision mod-w-line))
-(mod-w-test! "round-trip report-id" (get mod-w-back :report-id) "r1")
-(mod-w-test! "round-trip action" (get mod-w-back :action) "hide")
-(mod-w-test! "round-trip rule" (get mod-w-back :rule) "spam-hide")
-(mod-w-test! "round-trip tags :wire" (get mod-w-back :wire) true)
-(mod-w-test! "malformed → nil" (mod/wire->decision "garbage") nil)
-
-;; ── full federated transport: serialize → wire → deserialize → trust-gate ──
-
-(mod/fed-reset!)
-(define mod-w-peer-dec (mod/wire->decision mod-w-line))
-
-;; untrusted peer: decision is advisory, not applied
-(define mod-w-recv1 (mod/fed-receive-decision "peerX" mod-w-peer-dec))
-(mod-w-test!
-  "wired decision from untrusted peer → advisory"
-  (get mod-w-recv1 :applied)
-  false)
-(mod-w-test!
-  "untrusted wired decision not applied locally"
-  (mod/fed-applied-action "r1")
-  nil)
-
-;; trusted peer: decision binds locally
-(mod/grant-trust "peerY" :mod)
-(define mod-w-recv2 (mod/fed-receive-decision "peerY" mod-w-peer-dec))
-(mod-w-test!
-  "wired decision from trusted peer → applied"
-  (get mod-w-recv2 :applied)
-  true)
-(mod-w-test!
-  "trusted wired decision binds locally"
-  (get (mod/fed-applied-action "r1") :action)
-  "hide")
-
-(define mod-wire-tests-run! (fn () {:failures mod-w-failures :total mod-w-count :passed mod-w-pass :failed mod-w-fail}))

lib/mod/trace.sx

@@ -1,56 +0,0 @@
-;; lib/mod/trace.sx — policy dry-run diagnostics.
-;;
-;; decide-report returns the winning rule; a policy author debugging "why didn't
-;; my rule fire?" needs the whole picture. mod/trace-rules evaluates a report
-;; against every rule and reports each rule's proved/unproved status plus its
-;; goal-by-goal derivation — so an unproved rule shows exactly which goal failed.
-;; The winner is the first proved rule (same precedence as the engine).
-
-(define
-  mod/trace-rules
-  (fn
-    (r reports rules)
-    (let
-      ((count (mod/report-count (mod/report-about r) reports))
-        (id (mod/report-id r)))
-      (let
-        ((db (pl-load (mod/build-program r count rules))))
-        (let
-          ((proved-names (map (fn (s) (dict-get s "Rule")) (pl-query-all db (str "policy_action(" id ", _, Rule)")))))
-          (map
-            (fn (rule) (let ((nm (mod/rule-name rule))) {:proved (mod/member? nm proved-names) :goals (mod/proof-goals db id (mod/rule-when rule)) :action (mod/rule-action rule) :rule nm}))
-            rules))))))
-
-(define
-  mod/first-proved
-  (fn
-    (trace)
-    (reduce
-      (fn (acc t) (if (nil? acc) (if (get t :proved) t acc) acc))
-      nil
-      trace)))
-
-(define
-  mod/proved-rules
-  (fn
-    (trace)
-    (reduce
-      (fn
-        (acc t)
-        (if (get t :proved) (append acc (list (get t :rule))) acc))
-      (list)
-      trace)))
-
-(define
-  mod/trace-row
-  (fn
-    (t)
-    (str
-      (if (get t :proved) "[fires] " "[  -  ] ")
-      (get t :rule)
-      " → "
-      (get t :action))))
-
-(define
-  mod/trace-report
-  (fn (trace) (mod/join-with "\n" (map mod/trace-row trace))))

lib/mod/whatif.sx

@@ -1,56 +0,0 @@
-;; lib/mod/whatif.sx — policy what-if / impact analysis.
-;;
-;; Before shipping a policy change, a moderation team needs to know which past or
-;; pending reports would decide differently. mod/decision-diff compares one
-;; report's action under two rule sets; mod/policy-impact runs a whole batch and
-;; returns only the reports whose decision flips. Pure SX over decide-report.
-
-(define
-  mod/decision-diff
-  (fn
-    (r reports rules-a rules-b)
-    (let
-      ((a (get (mod/decide-report r reports rules-a) :action))
-        (b (get (mod/decide-report r reports rules-b) :action)))
-      {:after b :changed (if (= a b) false true) :report-id (mod/report-id r) :before a})))
-
-(define
-  mod/policy-impact
-  (fn
-    (reports rules-a rules-b)
-    (reduce
-      (fn
-        (acc r)
-        (let
-          ((d (mod/decision-diff r reports rules-a rules-b)))
-          (if (get d :changed) (append acc (list d)) acc)))
-      (list)
-      reports)))
-
-(define
-  mod/impact-count
-  (fn
-    (reports rules-a rules-b)
-    (len (mod/policy-impact reports rules-a rules-b))))
-
-(define
-  mod/impact-report
-  (fn
-    (reports rules-a rules-b)
-    (let
-      ((changed (mod/policy-impact reports rules-a rules-b)))
-      (if
-        (empty? changed)
-        "No decisions change."
-        (mod/join-with
-          "\n"
-          (map
-            (fn
-              (d)
-              (str
-                (get d :report-id)
-                ": "
-                (get d :before)
-                " → "
-                (get d :after)))
-            changed))))))

lib/mod/wire.sx

@@ -1,55 +0,0 @@
-;; lib/mod/wire.sx — portable decision wire format for federation transport.
-;;
-;; fed.sx shares decisions as in-memory dicts and leaves mod/fed-send! as the
-;; transport seam. This is the bytes that cross it: a versioned, pipe-delimited
-;; line encoding the verdict a peer needs (report id, action, rule) — enough to
-;; trust-gate and apply/advise, without shipping the whole proof tree. The
-;; loaded env has no string split, so split is built over slice/len.
-
-(define
-  mod/split-loop
-  (fn
-    (s ch n start pos acc)
-    (if
-      (= pos n)
-      (append acc (list (slice s start n)))
-      (if
-        (= (slice s pos (+ pos 1)) ch)
-        (mod/split-loop
-          s
-          ch
-          n
-          (+ pos 1)
-          (+ pos 1)
-          (append acc (list (slice s start pos))))
-        (mod/split-loop s ch n start (+ pos 1) acc)))))
-
-(define
-  mod/split-char
-  (fn (s ch) (mod/split-loop s ch (len s) 0 0 (list))))
-
-(define
-  mod/decision->wire
-  (fn
-    (d)
-    (str "MOD1|" (get d :report-id) "|" (get d :action) "|" (get d :rule))))
-
-(define
-  mod/wire-valid?
-  (fn
-    (w)
-    (let
-      ((parts (mod/split-char w "|")))
-      (if
-        (= (len parts) 4)
-        (= (nth parts 0) "MOD1")
-        false))))
-
-(define
-  mod/wire->decision
-  (fn
-    (w)
-    (if
-      (mod/wire-valid? w)
-      (let ((parts (mod/split-char w "|"))) {:action (nth parts 2) :wire true :rule (nth parts 3) :report-id (nth parts 1)})
-      nil)))

plans/agent-briefings/mod-loop.md

@@ -1,136 +0,0 @@
-# mod-on-sx loop agent (single agent, queue-driven)
-
-Role: iterates `plans/mod-on-sx.md` forever. **Moderation on Prolog** — reports,
-policy rules, decisions as backtracking proof search, audit trails, escalation
-state machine, federation. Where acl-sx asks "may this happen?", mod-sx asks
-"should this stay?" Sits on `lib/prolog/` (its test suite already green); adds a
-moderation-shaped vocabulary on top.
-
-```
-description: mod-on-sx queue loop
-subagent_type: general-purpose
-run_in_background: true
-isolation: worktree
-```
-
-## Prompt
-
-You are the sole background agent working `plans/mod-on-sx.md`. Isolated worktree
-`/root/rose-ash-loops/mod` on branch `loops/mod`, forever, one commit per feature.
-Push to `origin/loops/mod` after every commit. Never touch `main` or `architecture`.
-
-## Restart baseline — check before iterating
-
-1. Read `plans/mod-on-sx.md` — roadmap + Progress log.
-2. `ls lib/mod/` — pick up from the most advanced file.
-3. If `lib/mod/tests/*.sx` exist, run them via `bash lib/mod/conformance.sh`. Green
-   before new work.
-4. If `lib/mod/scoreboard.md` exists, that's your baseline.
-5. Read the `lib/prolog/` public API once — that's your substrate. The plan cites
-   `lib/prolog/prolog.sx` but that file does **not** exist; the real entry points
-   are `lib/prolog/runtime.sx`, `query.sx`, `compiler.sx`, `parser.sx`. Investigate
-   them (sx_find_all / grep for `(define ` heads) to learn how to assert facts and
-   run queries before writing any policy code.
-
-## The queue
-
-Phase order per `plans/mod-on-sx.md`:
-
-- **Phase 1** — report representation + simple policy (schema, defrule→clause,
-  `(decide id)` query, api). Tests: spam keyword → hide, repeated reports →
-  escalate, no rule → keep.
-- **Phase 2** — evidence accumulation + audit trail (proof tree from derivation,
-  append-only decision log, retrieval).
-- **Phase 3** — escalation + lifecycle state machine
-  (`:open → :triaged → :decided → :appealed → :final`), auto/human tiers, appeal.
-- **Phase 4** — federation (cross-instance reports, decision sharing, trust model,
-  revocation; mock fed-sx in tests).
-
-Within a phase, pick the checkbox that unlocks the most tests per effort.
-
-Every iteration: implement → test → commit → tick `[ ]` → Progress log → next.
-
-## Ground rules (hard)
-
-- **Scope:** only `lib/mod/**` and `plans/mod-on-sx.md`. Do **not** edit `spec/`,
-  `hosts/`, `shared/`, other `lib/<lang>/` dirs, `lib/stdlib.sx`, or `lib/` root.
-  May **import** from `lib/prolog/` only (its public API). Do **not** modify Prolog.
-- **NEVER call `sx_build`.** 600s watchdog. If the sx_server binary is broken →
-  Blockers entry, stop. Run tests by invoking the sx_server binary directly from a
-  conformance.sh (see how `lib/prolog/conformance.sh` drives it), pointing
-  `SX_SERVER` at `/root/rose-ash/hosts/ocaml/_build/default/bin/sx_server.exe`
-  (fresh worktrees have no `_build/`).
-- **Shared-file issues** → plan's Blockers with minimal repro; don't fix here.
-- **SX files:** `sx-tree` MCP tools ONLY. **They take `file:` not `path:`** — a
-  wrong key yields `Yojson Type_error("Expected string, got null")`, which looks
-  like a broken binary but is just a param mismatch. `sx_validate` after edits.
-  Path-based edits (`sx_replace_node`) count comment headers in their indices and
-  can clobber the wrong node — re-read after, or prefer `sx_write_file` for small
-  files. **Default to `sx_write_file` (rewrite the whole file) over path/pattern
-  edits** — these are small files and the rewrite always parses-before-writing.
-  `sx_insert_near` inserts only the FIRST top-level form of a multi-form source
-  (it silently drops the rest; byte count barely moves) — never use it to add a
-  block of forms; rewrite the file instead. `sx_replace_by_pattern` is fiddly to
-  match — don't fight it, just rewrite.
-- **Unicode in `.sx`:** raw UTF-8 only, never `\uXXXX` escapes.
-- **Commit granularity:** one feature per commit. Short factual messages
-  (`mod: spam-keyword policy rule → :hide + 6 tests`). Push to `origin/loops/mod`.
-- **Plan file:** update Progress log (newest first) + tick boxes every commit.
-
-## mod-specific gotchas
-
-- **Decisions are proofs, not booleans.** A decision should carry *why* — the
-  matching rule / derivation — so Phase 2's audit trail can persist it. Design the
-  Phase-1 `decide` return shape with that in mind (don't return a bare keyword you
-  later have to retrofit).
-- **Policy chains backtrack.** Order matters: first matching rule wins. Make rule
-  precedence explicit and deterministic (tests will depend on it). A "no rule
-  matched" outcome must be a real, testable result (`:keep`), not a query failure
-  you forget to handle.
-- **You may lean on backtracking and cut.** The substrate is full Prolog —
-  `pl-query-all` gives every proven clause (use it for "strictest-wins" or
-  multi-match analysis), `pl-query-one` gives the first (clause order = precedence).
-  Cut (`!`) and the other control constructs are available if you need to prune
-  alternatives inside a body, but for rule precedence prefer plain clause ordering
-  resolved by `pl-query-one` — it's the clean, testable default. Don't hand-roll
-  precedence in SX when the engine's backtracking already gives it to you.
-- **Negative decisions need closed-world care.** "No evidence of violation" vs
-  "evidence absent" differ. Be explicit about negation-as-failure where you use it.
-  In this substrate, negation is the **functor** `not(Goal)` / `\+(Goal)` — the
-  prefix `\+ Goal` operator does **not** parse. Unknown predicates *fail* (no
-  existence error), so a report lacking some fact safely falls through a rule that
-  references it. Quote user-data atoms (`'foo-bar'`) — a bare hyphen is the minus
-  operator and will misparse.
-- **Loaded-env strips the high-level string prims.** After the prolog preloads are
-  loaded, the eval env loses `includes?`, `chars`, `str-join`, `keyword` and
-  friends — they are **undefined** (a function calling one fails only when called,
-  often mid-test-load, looking like a mystery crash). Only the set the Prolog
-  tokenizer itself uses survives: `slice`, `len`, `nth`, `=`, `join` (sep first:
-  `(join sep list)`), `downcase`, `map`, `reduce`, `append`/`append!`, `when`,
-  `cond`, `if`, `let`, `begin`, `get`, `dict-get`, `keys`, `empty?`, `first`,
-  `reverse`, `+`, `-`, `<`, `<=`. Build substring search yourself over `slice`/
-  `len` (see `mod/str-contains?`). Treat `not`, `and`, `or`, `>` as suspect in
-  guest code unless you've confirmed them — nest `if`/`when` and use `(< a b)`.
-- **Lifecycle state is separate from policy.** Keep the state machine (Phase 3) as
-  an SX module over the engine, not tangled into Prolog rules.
-- **Federation trust is advisory by default.** A peer's decision only binds locally
-  when `(trust peer :mod)` holds; otherwise it's a suggestion. Don't auto-apply.
-
-## General gotchas (all loops)
-
-- SX `do` = R7RS iteration. Use `begin` for multi-expr sequences.
-- `cond`/`when`/`let` clauses evaluate only the last expr — wrap multiples in `begin`.
-- `let` is parallel, not sequential — nest `let`s when a binding references an earlier one.
-- `env-bind!` creates a binding; `env-set!` mutates an existing one (walks scope chain).
-- `sx_validate` after every structural edit.
-- Namespace-prefix all guest helpers (`mod/...`) — short/host-colliding names
-  (`bind`, `conj`, `name`) get silently shadowed or hang the runtime.
-
-## Style
-
-- No comments in `.sx` unless non-obvious.
-- No new planning docs — update `plans/mod-on-sx.md` inline.
-- Short, factual commit messages.
-- One feature per iteration. Commit. Log. Push. Next.
-
-Go. Start by reading the plan; find the first unchecked `[ ]`; implement it.

plans/flow-on-sx.md

@@ -16,7 +16,7 @@ federation extension via fed-sx for remote-node execution.
 
 ## Status (rolling)
 
-`bash lib/flow/conformance.sh` → **0/0** (not yet started)
+`bash lib/flow/conformance.sh` → **166/166** (Phases 1-8 complete; host ABI + reference driver)
 
 ## Ground rules
 
@@ -62,47 +62,167 @@ lib/flow/spec.sx           lib/flow/runtime.sx          lib/flow/store.sx
 
 ## Phase 1 — Declarative DAG + sequential execution
 
-- [ ] `lib/flow/spec.sx` — `defflow` macro, `sequence` combinator
-- [ ] node = Scheme thunk; output threads to next node (data flow)
-- [ ] `parallel` combinator (sequential semantics for now — TRUE parallelism in Phase 3)
-- [ ] runtime executes a flow synchronously, returns final value
-- [ ] `lib/flow/api.sx` — `(flow/start name args)` entry point
-- [ ] `lib/flow/tests/basic.sx` — 15+ cases: linear sequence, nested sequences,
-  data flow between nodes, parallel-with-join
-- [ ] `lib/flow/scoreboard.{json,md}`
-- [ ] `lib/flow/conformance.sh`
+- [x] `lib/flow/spec.sx` — `defflow` macro, `sequence` combinator
+- [x] node = Scheme procedure of one arg (upstream value threaded in); output
+  threads to next node (data flow). A node ignoring its arg is a thunk.
+- [x] `parallel` combinator (sequential semantics for now — TRUE parallelism in Phase 3)
+- [x] runtime executes a flow synchronously, returns final value
+- [x] `lib/flow/api.sx` — `(flow/start flow input)` entry point
+- [x] `lib/flow/tests/basic.sx` — 18 cases: single nodes, linear/nested sequence,
+  data flow between nodes, parallel-with-join, publish-shaped flow
+- [x] `lib/flow/scoreboard.{json,md}`
+- [x] `lib/flow/conformance.sh`
 
 ## Phase 2 — Control flow + error handling
 
-- [ ] `cond` combinator — predicate selects branch
-- [ ] `retry n [backoff]` — re-runs node up to n times on exception
-- [ ] `timeout ms` — bounds node execution
-- [ ] `try-catch` — exception handler with reified error
-- [ ] error model — exceptions vs explicit `(fail :reason ...)` results
-- [ ] `lib/flow/tests/control.sx` — 25+ cases: each combinator + composition
+- [x] `cond` combinator — predicate selects branch (named `branch`; `cond` is a
+  Scheme special form). `(branch pred then else)` — 6 tests.
+- [x] `retry n` — re-runs node up to n attempts on a raised exception; last
+  exception propagates. Only raised exceptions are retried — `(fail ...)` values
+  pass through. 6 tests. (Backoff deferred: no wall clock in pure SX.)
+- [x] `timeout budget` — bounds node execution via a **cooperative step budget**
+  (deterministic; no scheduler/clock in pure SX). Nodes opt in via `(tick)`;
+  `budget` ticks allowed, the next raises `flow-timeout`. Non-ticking nodes are
+  unbounded; budgets nest. 7 tests.
+- [x] `try-catch` — exception handler with reified error: `(try-catch node handler)`
+  runs node; on raise, calls `(handler error)` and returns its value. 6 tests.
+- [x] error model — exceptions vs explicit `(fail reason)` results: `fail`/`failed?`/
+  `fail-reason` produce/inspect failure values that flow downstream as data
+  (distinct from raised exceptions caught by retry/try-catch). 6 tests.
+- [x] `lib/flow/tests/control.sx` — 31 cases: branch, error model, try-catch,
+  retry, timeout + compositions
 
 ## Phase 3 — Suspend / resume (the showcase)
 
-- [ ] `(suspend reason)` — `call/cc` captures continuation, returns flow-id to caller
-- [ ] `lib/flow/store.sx` — serialize flow state (continuation + open vars)
-- [ ] `(flow/resume id value)` — load continuation, inject value, re-enter
-- [ ] `(flow/cancel id)` — explicit termination
-- [ ] crash recovery — on restart, scan store for paused flows, mark resumable
-- [ ] `lib/flow/tests/suspend.sx` — pause-resume scenarios, cancellation, "restart"
-  scenarios (simulated by re-loading store)
+- [x] `(suspend tag)` — guest call/cc is ESCAPE-ONLY (re-entry hangs), so resume
+  uses **deterministic replay**: suspend escapes to the driver as `(flow-suspended
+  tag)`; resume re-runs the flow, replaying resolved suspends from a `(tag value)`
+  log. No live continuation is ever serialized — the log is plain data.
+- [x] `lib/flow/store.sx` — flow store: id→record `(flow input log status payload)`;
+  `flow-drive` runs a flow against a replay log.
+- [x] `(flow/resume id value)` — append `(tag value)` to the log, re-drive; raw
+  result on completion, `(flow-suspended id tag)` on a further suspend.
+- [x] `(flow/cancel id)` — mark cancelled; a later resume is rejected (stale replay
+  cannot wake a cancelled flow).
+- [x] crash recovery — `flow-store-export` (procs nulled → plain data),
+  `flow-store-import!`, `flow-resumable-ids`. Records are name-keyed; resume
+  re-resolves the proc by name (defflow registers names), so a flow survives a
+  wiped store. `tests/recovery.sx`, 8 cases (export/wipe/import, resumable scan,
+  restart-at-every-step, replay-log survival).
+- [x] `lib/flow/tests/suspend.sx` — 17 cases: start/resume/cancel, multi-step,
+  replay determinism, lifecycle guards, suspend-in-branch
+- Harness: `flow-run` now reuses one env with a per-test reset (building the full
+  standard env 66× was too slow) — see `api.sx`.
 
 ## Phase 4 — Distributed nodes via fed-sx
 
-- [ ] `(remote-node addr fn args)` — execute node on a federation peer
-- [ ] failure semantics — retry on different peer, fall through to local
-- [ ] persistence across instances — flow state replicates via fed-sx
-- [ ] handoff — flow started here can resume on a peer if the local instance is down
-- [ ] `lib/flow/tests/distributed.sx` — federated flow scenarios (mock fed-sx in tests)
+- [x] `(remote-node addr fn)` — execute a node on a federation peer. Transport is
+  the fed-sx boundary, MOCKED via a peer registry (`flow-peer-register!`); raises
+  `flow-remote-unreachable` / `flow-remote-no-fn`. Composes with sequence, suspend,
+  retry. `tests/distributed.sx`, 7 cases.
+- [x] failure semantics — `(remote-failover addrs fn local)` tries each peer in
+  order, moves to the next on any raised error, and runs the `local` node if every
+  peer fails. 6 tests.
+- [x] persistence across instances — `(flow-replicate-to addr)` copies this
+  instance's store (the plain-data export) to a peer's replica slot;
+  `(flow-restore-from addr)` imports it. Same mechanism as crash recovery, across
+  instances.
+- [x] handoff — a flow started here resumes on a peer after the local instance dies:
+  replicate → wipe local store → restore on peer → `flow/resume`. The replay log
+  (and thus all resolved suspends) survives the move.
+- [x] `lib/flow/tests/distributed.sx` — 19 cases: remote-node, failover,
+  replication, handoff (including replay-log survival across the move)
+
+## Phase 5 — Operational API + combinator library
+
+The four roadmap phases are complete; this phase rounds out the engine into
+something operators and authors actually use. Accumulation, not a rewrite.
+
+- [x] introspection API — `flow/status id`, `flow/result id`, `flow/list`,
+  `flow/pending` (operator view of what each suspended flow awaits). 12 tests in
+  `tests/api.sx`.
+- [x] store hygiene — `flow/gc` drops terminal (done/cancelled) records keeping
+  live suspended flows (returns count); `flow/forget id` drops one terminal record
+  and refuses live flows. Bounds unbounded store growth. 9 tests in `tests/hygiene.sx`.
+- [x] `tap` — side-effecting pass-through node (logging/metrics) that returns input
+- [x] `recover` — complement to try-catch for the fail-VALUE channel: run node; if it
+  yields `(fail ...)`, run a recovery node on the reason
+- [x] `map-flow` — run a flow per item of a list, join results (sequential)
+- [x] `flow-while` / `flow-until` — bounded iteration: re-run body threading the
+  value while/until pred holds, capped at `max` steps (deterministic bound)
+- [x] `lib/flow/tests/api.sx` (12) + `lib/flow/tests/combinators.sx` (17)
+
+## Phase 6 — Railway-oriented composition
+
+Make the `(fail reason)` value channel compose into real validation/ETL pipelines.
+
+- [x] `attempt` — like `sequence`, but short-circuits at the first node that returns
+  a `(fail ...)` value, returning that failure (the railway track). Pairs with
+  `recover` for the rejoin.
+- [x] `lib/flow/tests/railway.sx` — 10 cases: fail short-circuiting, no-run-after-
+  failure, recover rejoin, validation pipeline reporting the failing stage
+
+## Phase 8 — Host integration ABI (art-dag / human-in-the-loop)
+
+`suspend` is the seam to the outside world, but a bare tag is an ad-hoc convention.
+This phase defines a stable request/response contract a host (an art-dag driver, a
+review UI) codes against — so flow can orchestrate art-dag with human decision
+points later without reverse-engineering tag shapes. `lib/flow/host.sx`.
+
+- [x] `(request kind payload)` — suspend with a typed `(flow-request kind payload)`
+  envelope; evaluates to the host's resume value. `await-human`/`await-render`/
+  `await-effect` sugar.
+- [x] `(flow-host-requests)` — the host work queue: `(id kind payload)` for every
+  suspended flow waiting on a host request; `request?`/`request-kind`/
+  `request-payload` parse a tag.
+- [x] `(flow-drive-host dispatch)` / `(flow-run-host dispatch maxticks)` — reference
+  host driver: the host supplies only a `(kind payload) -> answer` dispatch fn; the
+  loop drains pending requests and resumes until quiescent (bounded).
+- [x] `lib/flow/tests/host.sx` — 15 cases incl. the art-dag-shaped driver loop
+  (render → human-review → publish) run both manually and via `flow-run-host`.
+- Contract (documented in `host.sx` + README): the host owns IO + persistence; a
+  flow never does IO, it only `request`s; the host performs the effect and feeds the
+  result back via resume (logged, so not re-run on recovery). NOT done here (host
+  side, out of `lib/flow` scope): the real Celery/IPFS bridge and a persistent store
+  backend — those live in the art-dag integration, coding against this ABI.
+
+## Phase 7 — End-to-end integration
+
+Prove the phases compose: realistic flows exercising attempt + suspend + branch +
+remote-node + crash-recovery + handoff + introspection together.
+
+- [x] `lib/flow/tests/integration.sx` — 10 cases: an order-processing flow (validate
+  → payment suspend → branch → ledger federation) and an onboarding flow, run through
+  the full lifecycle including a simulated crash and a peer handoff mid-flow, plus
+  introspection (`flow/pending`/`status`/`result`) during the flow's life
 
 ## Progress log
 
-(loop fills this in)
+- **Phase 1 (combinators + sequential runtime).** Flow built as a Scheme prelude
+  loaded onto `scheme-standard-env`: a flow is a Scheme procedure `input -> output`,
+  so the whole flow runs inside the interpreter (sets up Phase 3 call/cc suspend).
+  Combinators `flow-node`/`flow-id`/`flow-const`/`sequence`/`parallel`/`defflow` in
+  `spec.sx`; `flow/start` + SX helpers (`flow-make-env`/`flow-run`) in `api.sx`.
+  18/18 in `tests/basic.sx`. Substrate constraints found: dotted rest params
+  `(a . rest)` and named `let` are unsupported in `lib/scheme/eval.sx`, so
+  combinators use `(lambda args ...)` variadics + top-level recursion. Scheme
+  strings come back boxed as `{:scm-string "..."}` — unwrap with `(get s :scm-string)`.
+
+- **Phases 2-4.** Control flow (branch/retry/timeout/try-catch + fail-value error
+  model), then the showcase: durable suspend/resume. Guest call/cc is escape-only
+  (re-entry hangs), so resume uses **deterministic replay** — re-run the flow,
+  replaying resolved suspends from a `(tag value)` log; only plain data persists, so
+  flows survive a wiped store (crash recovery) and a move to another instance
+  (replication + handoff). Phase 4 models the fed-sx boundary with a mock peer
+  registry. Timeout is a cooperative step budget (no wall clock in pure SX). Test
+  harness reuses one env with a per-test reset for speed.
+
+- **Phases 5-7 + docs.** Operational API (introspection, hygiene), combinator
+  library (tap/recover/map-flow/while/until), railway `attempt`, end-to-end
+  integration suite, and `lib/flow/README.md` (full API reference + replay-semantics
+  contract). **151/151 across 10 suites.** Conformance sx_server timeout raised to
+  540s for the 10-suite run under shared-machine CPU contention.
 
 ## Blockers
 
-(loop fills this in)
+(none)

plans/mod-on-sx.md

@@ -16,7 +16,7 @@ federation extension.
 
 ## Status (rolling)
 
-`bash lib/mod/conformance.sh` → **390/390** (roadmap + 19 extensions complete)
+`bash lib/mod/conformance.sh` → **0/0** (not yet started)
 
 ## Ground rules
 
@@ -66,373 +66,47 @@ lib/mod/fed.sx
 
 ## Phase 1 — Report representation + simple policy
 
-- [x] `lib/mod/schema.sx` — `report(id, by, about)`, `classification(id, kind)`,
-  `report_count(subject, n)` Prolog facts; keyword classifier derives evidence
-- [x] `lib/mod/policy.sx` — `mod/mk-rule` + ordered `mod/default-rules`; conditions
-  (`:classification`, `:count-at-least`) compile to Prolog goals; `policy_action/3`
-  clauses, last clause `true` so every report yields at least `:keep`
-- [x] `lib/mod/engine.sx` — `(mod/decide-report r reports rules)` queries
-  `policy_action(Id, Action, Rule)` with `pl-query-one` (clause order = precedence);
-  returns a decision dict `{:action :rule :report-id :proof}` carrying the why
-- [x] `lib/mod/api.sx` — registry + `(mod/report by about reason)`, `(mod/decide id)`
-- [x] `lib/mod/tests/decide.sx` — 31 cases: spam/abuse keyword, repeated→escalate,
-  no-rule→keep, precedence (spam beats repeated), proof shape, registry ids
-- [x] `lib/mod/scoreboard.{json,md}`
-- [x] `lib/mod/conformance.sh`
+- [ ] `lib/mod/schema.sx` — `report(id, by, about, reason)`, `evidence(id, kind, val)`,
+  `policy-action(report, action)` predicates as Prolog facts/rules
+- [ ] `lib/mod/policy.sx` — rule declarations: `(defrule action :when conditions)`
+  desugars to Prolog clause
+- [ ] `lib/mod/engine.sx` — `(decide report-id)` runs Prolog query, returns first
+  matching action
+- [ ] `lib/mod/api.sx` — `(mod/report by about reason)`, `(mod/decide id)`
+- [ ] `lib/mod/tests/decide.sx` — 15+ cases: spam keyword → hide, repeated reports →
+  escalate, no rule matches → keep
+- [ ] `lib/mod/scoreboard.{json,md}`
+- [ ] `lib/mod/conformance.sh`
 
 ## Phase 2 — Evidence + audit trail
 
-- [x] evidence accumulation — `report :evidence` list; `mod/attach-evidence` +
-  api `mod/add-evidence`; asserted as `evidence(Id, 'kind', 'val')` facts;
-  new `:evidence` condition + `reviewer-remove` rule consume it
-- [x] proof tree from Prolog derivation — `mod/proof-goals` re-queries each body
-  goal (id bound) against the same DB, recording goal text, solved?, and the
-  bindings that satisfied it (e.g. count goal yields N=3, S=subject)
-- [x] `lib/mod/audit.sx` — append-only log: monotonic `:seq`, decision + proof +
-  evidence snapshot; never mutates prior entries
-- [x] `(mod/audit id)` retrieval (+ `mod/audit-latest`, `mod/audit-all`, count)
-- [x] `lib/mod/tests/audit.sx` — 29 cases: proof goal text/bindings, evidence-driven
-  decisions, append-only ordering, per-report retrieval, snapshot-at-decision-time
+- [ ] evidence accumulation — additional facts asserted before query
+- [ ] proof tree from Prolog derivation tree
+- [ ] `lib/mod/audit.sx` — append-only log (decision + proof + evidence snapshot)
+- [ ] `(mod/audit id)` retrieval
+- [ ] `lib/mod/tests/audit.sx` — proof correctness, trail completeness
 
 ## Phase 3 — Escalation + lifecycle state machine
 
-- [x] state machine: `lib/mod/lifecycle.sx` — `:open → :triaged → :decided →
-  :appealed → :final` as a pure SX module over the engine; transition table guards
-  illegal moves (sets `:error`, leaves state); immutable cases with `:history`
-- [x] auto-tier: `mod/case-triage` runs the engine; terminal action (hide/remove/
-  keep) → tier `auto`, `mod/case-resolve` advances to `:decided`
-- [x] human-tier: `:escalate` action → tier `human`; `mod/case-resolve` is blocked
-  (sets `:error`); `mod/case-review` attaches evidence, re-decides, advances
-- [x] appeal: `mod/case-appeal` attaches appeal evidence + re-runs the engine; new
-  `exonerated-keep` rule (top precedence) lets exoneration override a prior `:hide`
-- [x] `(mod/appeal id new-evidence)` API — lifecycle façade over a case registry in
-  api.sx (`mod/triage` / `resolve` / `review` / `appeal` / `finalize`), logging
-  each committed decision to the audit trail
-- [x] `lib/mod/tests/escalation.sx` — 46 cases: transition guards, auto/human tiers,
-  blocked resolve, full appeal-override traversal, history, api façade
+- [ ] state machine: `:open → :triaged → :decided → :appealed → :final`
+- [ ] auto-tier: first-pass rules decide quick cases
+- [ ] human-tier: rules that emit `:escalate` move to next state
+- [ ] appeal: re-runs with appeal evidence, may override prior decision
+- [ ] `(mod/appeal id new-evidence)` API
+- [ ] `lib/mod/tests/escalation.sx` — full lifecycle traversal cases
 
 ## Phase 4 — Federation
 
-- [x] cross-instance reports — `mod/fed-receive-report peer …` ingests a peer's
-  report into the local registry, tagging origin; `mod/report-origin` resolves it
-  (local reports default to `"local"`); the engine decides federated reports
-  unchanged
-- [x] decision sharing — `mod/fed-share-decision decision peers` pushes messages to
-  the mock outbox (`mod/fed-send!` is the seam the real fed-sx transport replaces)
-- [x] trust model — `mod/fed-receive-decision` applies a peer's decision locally
-  ONLY when `(mod/trusted? peer :mod)`; otherwise it lands in the advisory log,
-  unapplied. `mod/grant-trust` / `mod/revoke-trust` manage the trust registry
-- [x] revocation — `mod/fed-revoke!` marks the applied action revoked + emits a
-  revocation message to the origin; `mod/fed-revoke-if-invalidated` re-runs the
-  engine and revokes only when the action no longer holds (proof invalidated)
-- [x] `lib/mod/tests/fed.sx` — 26 cases: trust grant/scope/revoke, cross-instance
-  ingest + origin, outbox sharing, advisory-vs-trusted apply, revocation +
-  invalidation (exoneration flips hide→keep → revoked)
-
-## Extensions (post-roadmap)
-
-- [x] **Ext 1 — negation-as-failure** (`lib/mod/tests/extensions.sx`, +14). Report
-  `:attrs`; policy conditions `(:attr "x")` → `attr(Id, x)` and `(:not <cond>)` →
-  `not(<cond>)` (the Prolog supports `not/1` and `\+/1` as *functors*, not the
-  prefix `\+` operator). Closed-world example: "hide spam UNLESS author verified".
-  Default policy untouched — demonstrated via custom rule sets, so all 132 base
-  tests stay green.
-- [x] **Ext 2 — weighted/aggregate scoring** (+8). Report `:signals` ({:kind
-  :weight}) project to `signal(Id, 'kind', weight)` facts; condition
-  `(:score-at-least N)` → `aggregate_all(sum(W), signal(Id, _, W), T), T >= N`.
-  Many weak signals accumulate past a threshold — genuine Prolog arithmetic
-  aggregation. Default policy untouched.
-- [x] **Ext 3 — proof explanation** (`lib/mod/explain.sx`, +10). `mod/explain`
-  renders a decision into a readable "why": action + rule, evidence line, and the
-  derivation goal-by-goal with `[proved]`/`[unproved]` marks and unification
-  bindings. E.g. `Report rc: escalate (rule: repeated-escalate)` … `[proved]
-  report(rc, B, S), report_count(S, N), N >= 3  {B=ann, N=3, S=dave}`.
-- [x] **Ext 19 — end-to-end triage pipeline** (`lib/mod/pipeline.sx`, +15).
-  `mod/triage-pipeline domain r reports actor` runs a report through domain-policy
-  decision → explanation → AP activity → wire, returning the full bundle. The test
-  is a genuine integration across 5 modules including a federated handoff (market
-  decision → wire → peer → trust-gated apply). The capstone that proves the
-  independently-built modules compose.
-- [x] **Ext 18 — ergonomic defrule / ruleset** (`lib/mod/defrule.sx`, +11). The
-  roadmap's `(defrule …)` surface, done with `&rest` variadics (no macro needed —
-  conditions are already plain data): `mod/defrule` collects trailing conditions,
-  `mod/ruleset` assembles rules. Produces structurally identical rules to `mk-rule`
-  and works in the engine unchanged.
-- [x] **Ext 17 — per-domain policy registry** (`lib/mod/policies.sx`, +14).
-  `mod/register-policy! domain rules` + `mod/decide-in domain r reports` give each
-  rose-ash domain (blog/market/events/…) its own rule set; unregistered domains
-  fall back to default-rules so a new domain is never unmoderated. Same spam report
-  → :remove under a strict market policy, :hide under blog's default.
-- [x] **Ext 16 — ActivityPub-shaped export** (`lib/mod/activity.sx`, +17).
-  `mod/decision->activity` maps a decision to a moderation verb (remove→Delete,
-  ban→Block, hide/escalate→Flag, keep→no activity) shaped like an AP activity
-  ({:type :actor :object :summary}), the precise mod action preserved in :action.
-  `mod/decisions->activities` batch-exports, dropping keeps — ready for the
-  platform's AP event bus / federated peers.
-- [x] **Ext 15 — disjunctive conditions** (`policy.sx` + `tests/disjunction.sx`,
-  +10). `(:any (list c1 c2 …))` compiles to Prolog disjunction `(g1 ; g2 ; …)`,
-  completing the condition boolean algebra (AND via the :when list, `:not`, `:any`).
-  Composes recursively — `:any` over `:not`/`:attr`/classification, and ANDs with
-  other conditions in the same rule. One rule now covers "spam OR abuse".
-- [x] **Ext 14 — decision wire format** (`lib/mod/wire.sx`, +16). The bytes that
-  cross `fed/fed-send!`: `mod/decision->wire` emits a versioned pipe-delimited line
-  (`MOD1|r1|hide|spam-hide`), `mod/wire->decision` parses it back (`mod/wire-valid?`
-  guards). Built `mod/split-char` over `slice`/`len` (loaded env has no split).
-  Integration test exercises the full path: serialize → wire → deserialize →
-  `fed-receive-decision` trust-gating (untrusted→advisory, trusted→applied).
-- [x] **Ext 13 — SLA sweep over pending cases** (`lib/mod/sla.sx`, +15). Composes
-  lifecycle (Phase 3) with time (Ext 12): a timed-case pairs a case with the tick
-  it entered its state; `mod/overdue?` flags pending cases (open/triaged/appealed)
-  past a deadline; `mod/sla-sweep` returns the breached report ids. Terminal states
-  never breach. Pure overlay — lifecycle stays timeless, the caller stamps entry.
-- [x] **Ext 12 — temporal burst detection** (`lib/mod/temporal.sx`, +15). Reports
-  gain an `:at` tick (deterministic, supplied — never clock-read).
-  `mod/decide-temporal now window` counts reports about the subject within
-  `[now-window, now]`, asserts `burst_count/2`, and a `(:burst-at-least K)` rule
-  fires only on a real burst. Verified: 3 reports at ticks 10/11/12 → hide;
-  3 reports at 1/2/12 (window 5) → keep, while the plain count rule escalates both.
-- [x] **Ext 11 — batch triage + corpus analytics** (`lib/mod/batch.sx`, +17).
-  `mod/decide-batch` triages a queue; `mod/action-histogram` summarizes outcomes by
-  action; `mod/rule-coverage` / `mod/never-fired` measure which rules fire across a
-  corpus — the *empirical* complement to lint's static unreachable check (Ext 5):
-  lint finds rules that can't fire, never-fired finds rules that didn't.
-- [x] **Ext 10 — policy what-if / impact** (`lib/mod/whatif.sx`, +13).
-  `mod/decision-diff` compares one report's action under two rule sets;
-  `mod/policy-impact` runs a batch and returns only the reports whose decision
-  flips; `mod/impact-count` / `mod/impact-report` summarize. Lets a team measure a
-  policy change before shipping it (e.g. "removing spam-hide flips r1 hide→keep").
-- [x] **Ext 9 — policy dry-run trace** (`lib/mod/trace.sx`, +15). `mod/trace-rules`
-  evaluates a report against every rule and returns each rule's proved/unproved
-  status + its goal-by-goal derivation, so an unproved rule shows which goal
-  failed. `mod/first-proved` = the winner (engine precedence), `mod/proved-rules`
-  the full firing set, `mod/trace-report` a `[fires]`/`[  -  ]` rendering. Answers
-  "why didn't my rule fire?" without instrumenting the engine.
-- [x] **Ext 8 — quorum over distinct reporters** (`lib/mod/quorum.sx`, +9). Anti-
-  brigade: `(:reporters-at-least N)` compiles to `setof(Br, report(_, Br, Sr), Bsr),
-  length(Bsr, Nr), Nr >= N` — distinct reporters, not raw report count.
-  `mod/decide-quorum` asserts every report's `report/3` fact (the base engine only
-  asserts the decided one) so Prolog can aggregate reporters. Verified one user
-  filing 3 reports stays `:keep` under quorum while the count rule would escalate.
-  (Substrate note: `^` existential doesn't parse; `setof(B, p(_, B, S), …)` with `_`
-  yields the distinct set in a single solution here.)
-- [x] **Ext 7 — repeat-offender escalation** (`lib/mod/offenders.sx`, +19). The
-  audit log as evidence: `mod/subject-sanctions` counts prior hide/remove/ban
-  decisions about a subject; `mod/decide-escalating id k` decides normally then
-  upgrades a *sanction* to `:ban` when the subject already has ≥k prior sanctions.
-  Non-sanction outcomes (keep/escalate) pass through untouched. First decision
-  whose input spans history beyond the single report — read from the trail, not
-  re-derived.
-- [x] **Ext 6 — strictest-wins strategy** (`lib/mod/severity.sx`, +14). Alternative
-  to first-match: `mod/decide-strictest` collects every proven rule (`pl-query-all`)
-  and picks the highest-`mod/action-severity` action (keep<escalate<hide<remove<ban).
-  Diverges from the default engine when rule order and severity disagree. Same
-  decision shape + `:strategy`; engine untouched.
-- [x] **Ext 5 — policy lint** (`lib/mod/lint.sx`, +14). Static analysis of a rule
-  set: `mod/unreachable-rules` flags rules placed after an unconditional (always-
-  matching) rule — structurally dead under first-match precedence;
-  `mod/has-catchall?` checks every report gets a decision; `mod/duplicate-rule-names`
-  + `mod/rules-ok?` give a one-call well-formedness verdict. No engine run needed.
-- [x] **Ext 4 — report linking / dedup** (`lib/mod/link.sx`, +12). `mod/related-ids`
-  and `mod/reporters-of` find reports about a subject via a Prolog relational query
-  (`report(Id, _, 'subject')`) — the policy substrate reused for retrieval.
-  `mod/dedup-reports` collapses identical reports (reporter|subject|reason key,
-  case-insensitive); `mod/distinct-reporters-of` counts unique reporters.
-
-## Shared-plumbing extraction — post-merge integration note
-
-mod-sx (Prolog) and acl-sx (Datalog, `lib/acl/`, 120/120) independently converged
-on the same module shape: `schema / engine / audit / explain / federation / api`.
-That parallel is the signal both plans flagged. **Recommendation: do NOT extract
-from a loop branch — extract at the architecture-merge integration point, after
-both `lib/mod` and `lib/acl` have landed, refactoring both consumers in one change.**
-
-- **Different engines.** acl = Datalog bottom-up (native derivation trees); mod =
-  Prolog backtracking (proof via per-goal `pl-query-all`). The engine and most of
-  `explain` are NOT shared — same intent, different mechanism. Don't try to unify them.
-- **Genuinely convergent shapes (the only real candidates):**
-  - **Append-only audit log** — `{seq, payload, retrieve-by-id}`; both have it (~40
-    lines). Lift to e.g. `lib/guest/audit-log.sx` parameterized by the entry payload.
-  - **Federation trust/outbox** — advisory-unless-`(trust peer :scope)` + a send
-    seam; both have it. Lift the trust registry + outbox; keep `:scope` a parameter
-    (`:mod` vs `:acl`).
-- **Trivia not worth a module:** `join-with`, `any?`, `str-contains?`, `distinct`.
-- **Why not now:** the branches merge independently; lifting from one leaves the
-  other's copy un-refactored → duplication, not sharing. Real extraction must touch
-  both consumers atomically, which only the post-merge integrator can do. Designing
-  the abstraction also needs both payload shapes in view (only mod's is visible here).
+- [ ] cross-instance reports — peer raises report about local content (or vice versa)
+- [ ] decision sharing — actions taken locally propagate to peers via fed-sx
+- [ ] trust model — peer's decision is advisory unless `(trust peer :mod)` is granted
+- [ ] revocation — undo applied moderation if proof was invalidated
+- [ ] `lib/mod/tests/fed.sx` — federated decision chains (mock fed-sx in tests)
 
 ## Progress log
 
-- **Ext 19 — end-to-end triage pipeline, 390/390** (+15). Capstone: one
-  orchestration call composes domain policy + decide + explain + activity + wire,
-  and the integration test runs the whole federated path (decide in a domain →
-  wire → peer → trust-gated apply) across 5 modules. Confirms the subsystem — built
-  module-by-module — actually composes end to end. mod-sx now spans schema → policy
-  DSL (boolean algebra + count/score/reporters/burst) → engine + proofs → audit →
-  lifecycle → SLA → federation (trust/wire/AP) → analytics (trace/whatif/lint/batch)
-  → domain policies → pipeline, all on the green lib/prolog substrate, 390 tests.
-- **Ext 18 — ergonomic defrule / ruleset, 375/375** (+11). Closes the roadmap's
-  original `defrule` surface. `fn` supports `&rest` here, and conditions evaluate
-  to plain data, so no macro is needed — variadic functions give the ergonomics
-  safely. Equivalence to `mk-rule` is asserted, so it's pure sugar with no new
-  semantics.
-- **Ext 17 — per-domain policy registry, 364/364** (+14). Multi-tenant policy:
-  the engine already took `rules` as a parameter, so domain-scoping is just a
-  registry + a default fallback — no engine change. Makes the whole policy
-  vocabulary (16 prior features) per-domain configurable. Default fallback means
-  adding a domain can't accidentally leave it unmoderated.
-- **Ext 16 — ActivityPub-shaped export, 350/350** (+17). Bridges mod-sx to the
-  wider rose-ash platform, which propagates cross-domain effects as AP-shaped
-  activities. Decisions become Flag/Delete/Block activities (keep = no-op); with
-  the wire format (Ext 14) and fed trust model (Phase 4) the federated moderation
-  path is now end-to-end: decide → activity/wire → peer → trust-gate → apply.
-- **Ext 15 — disjunctive conditions, 333/333** (+10). The condition DSL is now a
-  full boolean algebra: AND (the :when list), `:not` (NAF), `:any` (Prolog `;`).
-  cond->goal recurses, so the combinators nest arbitrarily — `:any` of `:not`s, an
-  `:any` ANDed with a `:not`, etc. — and the proof tree shows the compiled
-  disjunction verbatim. Maps directly onto Prolog's own control constructs rather
-  than reimplementing boolean logic in SX.
-- **Ext 14 — decision wire format, 323/323** (+16). Fills the federation transport
-  seam: decisions now serialize to a portable line and parse back, and the
-  integration test runs the whole federated path end-to-end (serialize on one
-  instance → trust-gated apply on another). Needed a hand-rolled `split-char`
-  (loaded env has no split) — over `slice`/`len`, same toolkit as `str-contains?`.
-- **Ext 13 — SLA sweep, 307/307** (+15). Two subsystems compose cleanly: lifecycle
-  states + temporal ticks → "which pending cases have sat too long". Kept lifecycle
-  pure by having the SLA layer carry entry-time externally (timed-case wrapper)
-  rather than stamping the case — same separation-of-concerns as keeping the state
-  machine out of Prolog.
-- **Ext 12 — temporal burst detection, 292/292** (+15). Adds the time dimension:
-  a windowed count distinguishes a burst from slow accumulation, where the plain
-  count rule cannot. Time is a supplied tick (`:at`), keeping everything
-  deterministic and testable — no clock primitive. Fifth report field (`:at`)
-  threaded through the rebuild helpers, same non-breaking pattern as
-  evidence/attrs/signals; all 277 prior tests stayed green.
-- **Ext 11 — batch triage + corpus analytics, 277/277** (+17). Operational layer:
-  triage a queue, histogram the outcomes, and measure rule coverage over real
-  data. `never-fired` pairs with lint (Ext 5) — static "can't fire" vs empirical
-  "didn't fire" — giving policy authors both views of dead rules. Histogram avoids
-  dict mutation by counting over a fixed action vocabulary.
-- **Ext 10 — policy what-if / impact, 260/260** (+13). Decisions are now
-  comparable across rule sets — diff one report, or batch a whole set and surface
-  only the flips. Pure SX over `decide-report`, no engine change. Closes the
-  policy-authoring loop alongside lint (Ext 5) and trace (Ext 9): lint checks
-  well-formedness, trace explains one report, what-if measures a change's blast
-  radius before it ships.
-- **Ext 9 — policy dry-run trace, 247/247** (+15). Whole-rule-set diagnostics over
-  the proof machinery: every rule's fire/no-fire and the goal that decided it. The
-  winner agrees with `decide-report` by construction (first proved = pl-query-one),
-  cross-checked in a test. Turns the proof tree from a per-decision artifact into a
-  policy-debugging tool.
-- **Ext 8 — quorum over distinct reporters, 232/232** (+9). Distinct-reporter
-  consensus via Prolog `setof`/`length`, requiring a second engine variant that
-  asserts all reports (the base engine deliberately scopes facts to the decided
-  report). Demonstrates the substrate handles set-aggregation, and that the
-  brigade case (one actor, many reports) is defeated by counting reporters not
-  reports. `^` existential doesn't parse here — `setof(B, p(_,B,S), …)` with `_`
-  gives the distinct set in one solution.
-- **Ext 7 — repeat-offender escalation, 223/223** (+19). Decisions can now depend
-  on history: the append-only audit log is read back as evidence, and a subject
-  with k prior sanctions has its next sanction upgraded to `:ban`. Closes the loop
-  between audit (Phase 2) and policy — the trail isn't just a record, it feeds
-  future decisions. Non-sanction outcomes never escalate (verified: a clean post
-  that the count rule escalates stays `:escalate`, never `:ban`).
-- **Ext 6 — strictest-wins strategy, 204/204** (+14). A second decision strategy
-  alongside first-match: collect all proven rules and apply the harshest sanction.
-  Shows the substrate supports more than one precedence policy over the same rule
-  facts — `pl-query-all` for the full match set, severity ranking in SX. Verified
-  it diverges from first-match exactly when rule order and severity disagree.
-- **Ext 5 — policy lint, 190/190** (+14). Static analysis of the rule set itself,
-  catching the failure modes first-match precedence makes easy: dead rules after a
-  catch-all, missing catch-all (undecided reports), duplicate names. `mod/rules-ok?`
-  is a single well-formedness gate a policy author can assert in their own tests.
-- **Ext 4 — report linking / dedup, 176/176** (+12). Relational retrieval
-  (`related-ids`, `reporters-of`) reuses the Prolog substrate for *querying* report
-  clusters, not just deciding them — `report(Id, _, 'subject')` by unification.
-  Dedup is pure SX over a normalized link key. Own suite (`tests/link.sx`) — going
-  forward, new extensions get their own test file rather than growing
-  `extensions.sx`. With roadmap + 4 extensions the subsystem now spans schema →
-  policy DSL (6 condition types) → engine + proofs → audit → lifecycle →
-  federation → explanation → linking, all on the green `lib/prolog` substrate.
-- **Ext 3 — proof explanation, 164/164** (+10). `mod/explain` turns the Phase-2
-  proof tree into human-readable text — the audit trail's "why" made legible. Pure
-  SX over existing decision data; no engine change. Renders unification bindings
-  inline (`{B=ann, N=3, S=dave}`) so a moderator sees exactly which facts proved
-  the decision.
-- **Ext 2 — weighted/aggregate scoring, 154/154** (+8). `:signals` + the
-  `(:score-at-least N)` condition push aggregation into Prolog
-  (`aggregate_all(sum(W), …)`), so low-confidence signals can accumulate to a
-  takedown. The schema's report-rebuild helpers (`report*` / `with-*`) now thread
-  six fields; each addition stays non-breaking because empty collections project
-  to empty fact blocks. Default policy and its 132 tests untouched (proven via
-  custom rule sets).
-- **Ext 1 — negation-as-failure, 146/146** (+14). `:attr` and `:not` conditions
-  give the policy closed-world reasoning. The substrate's negation is a functor
-  (`not(Goal)`), not the ISO prefix `\+` operator (that doesn't parse here) —
-  noted for any future negation work. Kept the default rule set and its 132 tests
-  untouched by proving the feature through custom rule sets instead.
-- **Phase 4 complete — 132/132** (+26 fed). **Full roadmap done.** Federation:
-  cross-instance reports, decision sharing, advisory-by-default trust, revocation.
-  fed-sx is mocked behind `mod/fed-send!` (in-memory outbox) — the only seam a real
-  transport must replace. The hard rule is enforced: a peer's decision binds
-  locally only under `(mod/trusted? peer :mod)`; otherwise it is recorded as a
-  suggestion and never auto-applied. Revocation composes with the proof model from
-  Phase 2 — `mod/fed-revoke-if-invalidated` re-runs the *same* engine and undoes a
-  moderation only when the action it once proved no longer holds (an exoneration
-  evidence flips hide→keep, triggering revocation + an origin-bound revocation
-  message).
-  - **Liftable (acl-sx watch):** the trust registry (`grant`/`revoke`/`trusted?`
-    over `{:peer :scope}`) and the outbox/send! seam are generic federation
-    plumbing; candidates for `lib/guest/` if acl-sx grows a federation phase.
-- **Phase 3 complete — 106/106** (+46 escalation). Lifecycle state machine,
-  auto/human tiers, appeal-override, and an api façade. The state machine is a
-  pure SX module (`lib/mod/lifecycle.sx`) over the engine — policy stays in
-  Prolog, lifecycle stays out of it, per the design constraint. Cases are
-  immutable values threaded through transitions; illegal moves set `:error`
-  rather than throwing (the env's error handling is untested, so this keeps tests
-  deterministic). Tier logic: triage runs the engine, an `:escalate` action parks
-  the case at the human tier where `mod/case-resolve` is blocked until
-  `mod/case-review` supplies evidence. Appeal-override works because the new
-  `exonerated-keep` rule sits at top precedence — appeal evidence re-runs the same
-  engine and a higher-precedence clause wins. The api façade (`mod/triage` …
-  `mod/finalize`) keeps a per-report case registry and logs each committed
-  decision to the Phase-2 audit trail, so lifecycle + audit compose.
-  - **Gotcha:** `sx_insert_near` inserts only the FIRST top-level form of a
-    multi-form source — silently drops the rest (byte count barely changes). For
-    multi-form additions, rewrite the file with `sx_write_file`.
-- **Phase 2 complete — 60/60** (+29 audit). Evidence accumulation, constructive
-  proof trees, append-only audit log. A decision's `:proof :goals` is a real
-  derivation: each body goal is re-queried against the same Prolog DB with the
-  report id bound, so the count rule's proof carries `N=3, S=<subject>` straight
-  from unification — not a reconstruction. Evidence is asserted as
-  `evidence(Id, 'kind', 'val')`; the new `reviewer-remove` rule (placed first =
-  highest precedence) lets human review override automated classification.
-  `mod/decide` now commits each decision to the audit log with the evidence
-  snapshot in force at decision time. Unknown predicates in this Prolog fail
-  gracefully (verified) — so an evidence-less report safely falls through the
-  reviewer rule without an existence error.
-  - **Liftable (acl-sx watch):** the proof-tree builder (`mod/proof-goals` —
-    re-query-each-goal) and the append-only log shape are both generic. Both
-    subsystems are now past Phase 2; next time either touches plumbing, evaluate
-    lifting `proof-goals` + the audit-log primitives into `lib/guest/`.
-- **Phase 1 complete — 31/31.** Report schema, keyword classifier, policy DSL,
-  engine, registry api, conformance harness. Decisions are proofs: each carries
-  `:rule` (matching clause), `:proof {:rule :conditions :evidence :count}`.
-  Precedence is Prolog clause order resolved by `pl-query-one`; a trailing
-  `true`-bodied default rule makes "no rule matched" a real `:keep`, not a query
-  failure. Evidence (spam/abuse classification) derived in SX and asserted as
-  `classification/2` facts; repeated-report escalation uses a genuine Prolog
-  join + arithmetic (`report(Id,_,S), report_count(S,N), N >= 3`).
-  - **Gotcha (env):** loading the prolog libs strips `includes?` (and other
-    high-level string prims) from the eval env — only the set the prolog
-    tokenizer itself uses survives (`slice`, `len`, `nth`, `=`, `join`,
-    `downcase`, `map`, `reduce`, `append!`). Implemented `mod/str-contains?` over
-    `slice`/`len` rather than relying on `includes?`. Watch for this in later
-    phases — stick to the blessed primitive set.
-  - **Liftable (acl-sx watch):** `mod/join-with`, `mod/str-contains?`, `mod/any?`,
-    and the rule→clause compilation shape are generic rule-engine plumbing. Do not
-    extract to `lib/guest/` until both mod-sx and acl-sx are past Phase 2.
+(loop fills this in)
 
 ## Blockers
 
-(none)
+(loop fills this in)