--- name: memory description: Query and store learnings in the persistent memory system user-invocable: true --- # Memory System Unified interface for storing and retrieving learnings across sessions. ## When to Use - "What did we do before with X?" - "Remember this for next time" - Starting work similar to past sessions - Looking for patterns that worked\/failed - Debugging recurring issues --- ## Recall (Query Memory) ### Quick Usage ``` \/recall ``` ### Examples ```bash # Semantic search \/recall hook development patterns \/recall TypeScript errors \/recall wizard installation # With options \/recall hook patterns --k 10 --vector-only ``` ### Execution ```bash cd $CLAUDE_PROJECT_DIR\/opc && PYTHONPATH=. uv run python scripts\/core\/recall_learnings.py --query "" --k 5 ``` ### Options | Flag | Description | |------|-------------| | `--k N` | Return N results (default: 5) | | `--vector-only` | Pure vector search (higher precision) | | `--text-only` | Text search only (faster) | ### Output Format ``` ## Memory Recall: "" ### 1. [TYPE] (confidence: high, id: abc123) ### 2. [TYPE] (confidence: medium, id: def456) ``` --- ## Remember (Store Learning) ### Quick Usage ``` \/remember ``` ### With Explicit Type ``` \/remember --type WORKING_SOLUTION ``` ### Examples ```bash \/remember TypeScript hooks require npm install before they work \/remember --type ARCHITECTURAL_DECISION Session affinity uses terminal PID \/remember --type FAILED_APPROACH Don't use subshell for store_learning command ``` ### Learning Types | Type | Use For | |------|---------| | `WORKING_SOLUTION` | Fixes, solutions that worked (default) | | `ARCHITECTURAL_DECISION` | Design choices, system structure | | `CODEBASE_PATTERN` | Patterns discovered in code | | `FAILED_APPROACH` | What didn't work | | `ERROR_FIX` | Specific error resolutions | | `USER_PREFERENCE` | User's preferred approaches | | `OPEN_THREAD` | Incomplete work to resume | ### Execution ```bash cd $CLAUDE_PROJECT_DIR\/opc && PYTHONPATH=. uv run python scripts\/core\/store_learning.py \ --session-id "manual-$(date +%Y%m%d-%H%M)" \ --type \ --content "" \ --context "manual entry via \/remember" \ --confidence medium ``` ### Auto-Type Detection If no `--type` specified, infer from content: - Contains "error", "fix", "bug" → ERROR_FIX - Contains "decided", "chose", "architecture" → ARCHITECTURAL_DECISION - Contains "pattern", "always", "convention" → CODEBASE_PATTERN - Contains "failed", "didn't work", "don't" → FAILED_APPROACH - Default → WORKING_SOLUTION --- ## Recall Reasoning (Artifact Index) Search artifact index for handoffs, plans, and post-mortems. ### When to Use - Find what worked\/failed in past sessions - Look up architectural decisions - Review post-mortems from completed work ### Usage ```bash uv run python scripts\/artifact_query.py "" [--outcome SUCCEEDED|FAILED] [--limit N] ``` ### Examples ```bash # Search for auth work uv run python scripts\/artifact_query.py "authentication OAuth" # Only successful approaches uv run python scripts\/artifact_query.py "implement agent" --outcome SUCCEEDED # What failed (avoid repeating) uv run python scripts\/artifact_query.py "hook implementation" --outcome FAILED ``` ### Interpreting Results - `✓` = SUCCEEDED (follow this pattern) - `✗` = FAILED (avoid this pattern) - `?` = UNKNOWN (not yet marked) --- ## Memory Architecture | Component | Purpose | |-----------|---------| | PostgreSQL | Primary storage with pgvector | | BGE Embeddings | 1024-dim vectors (bge-large-en-v1.5) | | Hybrid Search | RRF combining text + vector | | Artifact Index | Handoffs\/plans with post-mortems | ### Score Interpretation | Search Mode | Score Range | Meaning | |-------------|-------------|---------| | Hybrid RRF (default) | 0.01-0.03 | Good (RRF ranking) | | Vector-only | 0.4-0.6 | Cosine similarity | | Text-only | 0.01-0.05 | BM25 normalized | **Note:** Low RRF scores (0.02) are GOOD - it's a ranking fusion metric. --- ## Proactive Memory Usage ### Before Starting Work ```bash \/recall ``` ### After Solving Problems ```bash \/remember ``` ### Before Similar Tasks Check artifact index for past approaches that succeeded or failed.