3 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
poimenandrock	d7a36ce9e8	ci: optimize build + deploy + migrate workflows (#51) ... ## Optimize CI/CD Workflows ### Changes #### build.yaml - **Merge 3 cargo steps → 1 compile pass**: `cargo build`, `cargo test`, `cargo clippy` now run in single invocation, reusing compiled artifacts - **Remove `cargo clean`**: Eliminated wasteful step that deleted artifacts before Docker build - **Add secret validation**: Registry credentials checked before login (fail-fast) #### deploy.yaml - **Skip checkout**: Removed unnecessary git clone - **Fetch SHA via Gitea API**: Query latest commit directly instead of cloning - **Reuse existing token**: Use `FORGEJO_REGISTRY_TOKEN` for Gitea API auth (already has privileges) - **Validate image exists**: Check SHA image exists before tagging as latest (prevents tagging non-existent images) - **Add secret validation**: Registry credentials checked before login (fail-fast) #### migrate.yaml - **Merge schema verification**: Schema inspect result reused in both changed + manual paths - **Fix manual trigger errors**: Manual mode now fails on first migration error (was silently masking with `|| true`) - **Track failures**: Explicit FAILED flag tracks migration errors across loop ### Benefits - **Speed**: Fewer compiles, no unnecessary clones, reuse artifacts - **Reliability**: Secret validation catches configuration issues early - **Safety**: Image existence check prevents tagging phantom images - **Clarity**: Merged steps have descriptive names, explicit error handling ### Testing - Branch: `ci/optimize-workflows` - Ready to merge to `main` after review --------- Co-authored-by: rock <[email protected]> Reviewed-on: #51 Co-authored-by: poimen <[email protected]>	2026-09-13 05:42:01 +00:00
poimenandrock	9f70109c1d	feat: scale memory-db to 3 replicas for HA (#50) ... ✅ All 3 replicas running and synced - memory-db-1 (primary) - memory-db-2 (replica, LSN 0/9000060) - memory-db-3 (replica, LSN 0/9000060) Cluster status: healthy Production-ready for failover. --------- Co-authored-by: rock <[email protected]> Reviewed-on: #50 Co-authored-by: poimen <[email protected]>	2026-09-13 00:04:32 +00:00
poimenandrock	fb61de6b47	feat: LLM entity + fact extraction pipeline (Zep paper alignment) (#48) ... ## Changes ### Entity Extraction - Switch from WikiLinkFallbackExtractor to LlmEntityExtractor when LLM_ENDPOINT set - `clean_llm_response()`: strips `<think>` tags, markdown fences, extracts JSON - Handle array responses (Ollama returns `[...]` not `{entities: [...]}`) - EntityType custom Deserialize: unknown variants → Unknown (no crash) - Increase timeout 30s→90s, max_tokens 500→1500 for reasoning models - Graceful reflection fallback: keep entities if verification fails ### Fact Extraction (NEW) - LlmFactExtractor: LLM-based relationship extraction between entity pairs - Validates source/target against known entity list (drops hallucinated edges) - Same robust JSON cleaning for reasoning models + Ollama - IngestWorker auto-selects LLM vs Simple based on LLM_ENDPOINT env ### K8s Deployment - Add `command: ["/app/mem"]` (fix args replacing CMD) - Add LLM_ENDPOINT, LLM_MODEL env vars for in-cluster LLM ## E2E Tested (local Ollama qwen2.5:3b) - 12 entities extracted (person, tool, concept, organization) - 5 edges with relationships and facts - 781 tests pass ## Zep Paper Alignment (§2.2) - Entity extraction + resolution (§2.2.1) - Fact extraction between entity pairs (§2.2.2) - Temporal edge invalidation ready (t_valid/t_invalid schema) - Reflection verification (§2.2.1, graceful fallback) --------- Co-authored-by: rock <[email protected]> Reviewed-on: #48 Co-authored-by: poimen <[email protected]>	2026-09-11 01:11:15 +00:00