- Update save_run_cache to also update actor_id, recipe, inputs on conflict
- Add logging for actor_id when saving runs to run_cache
- Add admin endpoint DELETE /runs/admin/purge-failed to delete all failed runs
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Use dedicated thread with new event loop for database operations
- Create new database connection per operation to avoid pool conflicts
- Handles both async and sync calling contexts correctly
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Database is the ONLY source of truth for cache_id -> ipfs_cid
- Removed Redis caching layer entirely
- Failures will raise exceptions instead of warning and continuing
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Database (cache_items table) is now source of truth
- Redis used as fast cache on top
- Mapping persists across restarts
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Files in /data/cache/nodes/ are now stored by IPFS CID only
- cache_id parameter creates index from cache_id -> IPFS CID
- Removed deprecated node_id parameter behavior
- get_by_cid(cache_id) still works via index lookup
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
IPFS CIDs are the primary identifiers. If IPFS upload fails,
the operation must fail rather than silently using local hashes.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add COMPOUND node handling in execute_recipe for collapsed effect chains
- Index cache entries by node_id (cache_id) when different from IPFS CID
- Fix test_cache_manager.py to unpack put() tuple returns
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Compiler now generates SHA3-256 hashes for node IDs
- Each hash includes type, config, and input hashes (Merkle tree)
- Same plan = same hashes = automatic cache reuse
Cache changes:
- Remove index.json - filesystem IS the index
- Files at {cache_dir}/{hash}/output.* are source of truth
- Per-node metadata.json for optional stats (not an index)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When content isn't found in local cache, fetch directly from IPFS
using the CID. IPFS is the source of truth for all content-addressed data.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The artdag Cache object doesn't persist state across process restarts,
so cache.get(node_id) returns None even when files exist on disk.
Now we check the filesystem directly at {cache_dir}/nodes/{node_id}/output.*
when the in-memory cache lookup fails but we have a valid node_id from
the Redis index.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Debug why recipes are not found in cache after upload.
Logs now show each step of put() and get_by_cid().
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- cache_service.delete_content: Remove user's ownership link first,
only delete actual file if no other owners remain
- cache_manager.discard_activity_outputs_only: Check if outputs and
intermediates are used by other activities before deleting
- run_service.discard_run: Now cleans up run outputs/intermediates
(only if not shared by other runs)
- home.py clear_user_data: Use ownership model for effects and media
deletion instead of directly deleting files
The ownership model ensures:
1. Multiple users can "own" the same cached content
2. Deleting removes the user's ownership link (item_types entry)
3. Actual files only deleted when no owners remain (garbage collection)
4. Shared intermediates between runs are preserved
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The dual-indexing code was calling file_hash(source_path) after
cache.put(move=True) had already moved the file, causing
"No such file or directory" errors on upload.
Now computes local_hash before the move operation.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add friendly name display to media detail and list pages
- Unpack nested meta fields to top level for template access
- Fix output_cid mismatch: use IPFS CID consistently between cache and database
- Add dual-indexing in cache_manager to map both IPFS CID and local hash
- Fix plan display: accept IPFS CIDs (Qm..., bafy...) not just 64-char hashes
- Add friendly names to recipe listing
- Add recipe upload button and handler to recipes list
- Add debug logging to recipe listing
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fix list_by_type to return node_id (IPFS CID) instead of local hash
- Fix effects count on home page (count from _effects/ directory)
- Add nav_counts to all page templates (recipes, effects, runs, media, storage)
- Add editable metadata section to cache/media detail page
- Show more metadata on recipe detail page (ID, IPFS CID, step count)
- Update tests for new list_by_type behavior
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Bug: Two get_by_cid methods existed in L1CacheManager. The second
definition shadowed the first, breaking recipe lookup because the
comprehensive method (using find_by_cid) was hidden.
- Remove duplicate get_by_cid method (lines 470-494)
- Add regression test to ensure only one get_by_cid exists
- Add tests for template variables and recipe visibility
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Refactor to use IPFS CID as the primary content identifier:
- Update database schema: content_hash -> cid, output_hash -> output_cid
- Update all services, routers, and tasks to use cid terminology
- Update HTML templates to display CID instead of hash
- Update cache_manager parameter names
- Update README documentation
This completes the transition to CID-only content addressing.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Upload endpoint returns both CID and content_hash
- Cache manager handles both SHA3-256 hashes and IPFS CIDs
- get_by_cid() fetches from IPFS if not cached locally
- Execute tasks support :cid in addition to :hash
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Recipes: Now content-addressed only (cache + IPFS), removed Redis storage
- Runs: Completed runs stored in PostgreSQL, Redis only for task_id mapping
- Add list_runs_by_actor() to database.py for paginated run queries
- Add list_by_type() to cache_manager for filtering by node_type
- Fix upload endpoint to return size and filename fields
- Fix recipe run endpoint with proper DAG input binding
- Fix get_run_service() dependency to pass database module
Storage architecture:
- Redis: Ephemeral only (sessions, task mappings with TTL)
- PostgreSQL: Permanent records (completed runs, metadata)
- Cache: Content-addressed files (recipes, media, outputs)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The get_cache_manager() singleton wasn't initializing with Redis,
so workers couldn't see files uploaded via the API server.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The cache_manager now uses Redis hashes for the content_index and
ipfs_cids mappings. This allows multiple uvicorn workers to share
state, so files added by one worker are immediately visible to all
others.
- Added redis_client parameter to L1CacheManager
- Index lookups check Redis first, then fall back to in-memory
- Index updates go to both Redis and JSON file (backup)
- Migrates existing JSON indexes to Redis on first load
- Re-enabled workers=4 in uvicorn
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add PostgreSQL database for cache metadata storage with schema for
cache_items, item_types, pin_reasons, and l2_shares tables
- Add IPFS integration as durable backing store (local cache as hot storage)
- Add postgres and ipfs services to docker-compose.yml
- Update cache_manager to upload to IPFS and track CIDs
- Rename all config references to recipe throughout server.py
- Update API endpoints: /configs/* -> /recipes/*
- Update models: ConfigStatus -> RecipeStatus, ConfigRunRequest -> RecipeRunRequest
- Update UI tabs and pages to show Recipes instead of Configs
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add PyYAML dependency for parsing config files
- Add Pydantic models: VariableInput, FixedInput, ConfigStatus, ConfigRunRequest
- Add Redis storage functions for configs
- Add config YAML parsing with variable and fixed input detection
- Add config API endpoints: upload, list, get, delete, run
- Add config UI: Configs tab, list page, detail page with run form
- Add HTMX endpoints for config operations
- Add pinning on publish: configs and their fixed inputs are pinned
when runs from configs are published to L2
- Clean up debug logging in cache_manager
Config YAML format supports:
- Fixed inputs: resolve asset hashes from registry
- Variable inputs: marked with `input: true`, filled at run time
- DAG definition with nodes and edges
- Registry of assets and effects
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add is_pinned(), pin(), _load_meta(), _save_meta() to L1CacheManager
- Update can_delete() and can_discard_activity() to check local pinned status
- Update run deletion endpoints (API and UI) to check pinned metadata
- Remove L2 shared check fallback from run deletion
- Fix L2SharedChecker to return True on error (safer - prevents accidental deletion)
- Update tests for new pinned behavior
When items are published to L2, the publish flow marks them as pinned
locally. This ensures items remain non-deletable even if L2 is unreachable,
and both outputs AND inputs of published runs are protected.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
cache_manager.py:
- get_by_content_hash() now tries direct cache.get(content_hash)
since uploads use content_hash as node_id
- This works even if cache index hasn't been reloaded
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Update list_all() to scan cache_dir for legacy files directly
(old files stored as CACHE_DIR/{hash}, not CACHE_DIR/legacy/)
- Update cache listing endpoints to use cache_manager.list_all()
instead of iterating CACHE_DIR.iterdir() directly
- This ensures uploaded files appear in the cache UI regardless
of whether they're in the old or new cache structure
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove symlink hack from cache_file() - no longer needed
- Add get_cache_path() helper for content_hash lookups
- Update all CACHE_DIR / content_hash patterns to use cache_manager
- Fix cache_manager.get_by_content_hash() to check path.exists()
- Fix legacy path lookup (cache_dir not legacy_dir)
- Update upload endpoint to use cache_manager.put()
This ensures cache lookups work correctly for both legacy files
(stored directly in CACHE_DIR) and new files (stored in nodes/).
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add cache_manager.py with L1CacheManager wrapping artdag Cache
- Add L2SharedChecker for checking published status via L2 API
- Update server.py to use cache_manager for storage
- Update DELETE /cache/{content_hash} to enforce deletion rules
- Add DELETE /runs/{run_id} endpoint for discarding runs
- Record activities when runs complete for deletion tracking
- Add comprehensive tests for cache manager
Deletion rules enforced:
- Cannot delete items published to L2
- Cannot delete inputs/outputs of runs
- Can delete orphaned items
- Runs can only be discarded if no items are shared
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
