HN 표시: 생물학적 부패가 포함된 AI 메모리(52% 재현율)

Every session, your AI assistant starts from zero. It asks the same questions, forgets your preferences, re-learns your stack. There is no memory between conversations. YourMemory fixes that. It gives AI agents a persistent memory layer that works the way human memory does — important things stick, forgotten things fade, outdated facts get replaced automatically. Two commands to install, zero infrastructure required. Tested on LoCoMo-10 — 1,534 QA pairs across 10 multi-session conversations. | System | Recall@5 | 95% CI | |---|---|---| | YourMemory (BM25 + vector + graph + decay) | 59% | 56–61% | | Zep Cloud | 28% | 26–30% | 2× better recall than Zep Cloud on the same benchmark. Full methodology and per-sample breakdown in BENCHMARKS.md. Writeup: I built memory decay for AI agents using the Ebbinghaus forgetting curve. Supports Python 3.11, 3.12, 3.13, and 3.14. No Docker, no database setup, no external services. pip install yourmemory yourmemory-setup Downloads the spaCy language model and initialises the local database at ~/.yourmemory/memories.duckdb . yourmemory-path Prints your full executable path and a ready-to-paste config block. Copy it. Claude Code Add to ~/.claude/settings.json : { "mcpServers": { "yourmemory": { "command": "yourmemory" } } } Reload (Cmd+Shift+P → Developer: Reload Window ). Claude Desktop Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows): { "mcpServers": { "yourmemory": { "command": "yourmemory" } } } Restart Claude Desktop. Cline (VS Code) VS Code doesn't inherit your shell PATH. Run yourmemory-path first to get the full executable path. In Cline → MCP Servers → Edit MCP Settings: { "mcpServers": { "yourmemory": { "command": "/full/path/to/yourmemory", "args": [], "env": { "YOURMEMORY_USER": "your_name" } } } } Restart Cline after saving. Cursor Add to ~/.cursor/mcp.json : { "mcpServers": { "yourmemory": { "command": "/full/path/to/yourmemory", "args": [], "env": { "YOURMEMORY_USER": "your_name" } } } } OpenCode Add to ~/.config/opencode/config.json : { "mcp": { "yourmemory": { "type": "local", "command": ["yourmemory"], "environment": { "YOURMEMORY_USER": "your_name" } } } } Then copy the memory workflow instructions: cp sample_CLAUDE.md ~/.config/opencode/instructions.md Restart OpenCode. Any MCP-compatible client: YourMemory is a standard stdio MCP server. Works with Windsurf, Continue, Zed, and any client that supports MCP. Use the full path from yourmemory-path if the client doesn't inherit shell PATH. cp sample_CLAUDE.md CLAUDE.md Edit CLAUDE.md — replace YOUR_NAME and YOUR_USER_ID . Claude now follows the recall → store → update workflow automatically on every task. Three tools. Called by Claude automatically once CLAUDE.md is in place. | Tool | When | What it does | |---|---|---| recall_memory(query) | Start of every task | Surfaces relevant memories ranked by similarity × strength | store_memory(content, importance) | After learning something new | Embeds and stores with biological decay | update_memory(id, new_content) | When a memory is outdated | Re-embeds and replaces | # Example session store_memory("Sachit prefers tabs over spaces in Python", importance=0.9, category="fact") # Next session — without being told again: recall_memory("Python formatting") # → {"content": "Sachit prefers tabs over spaces in Python", "strength": 0.87} | Category | Survives without recall | Use case | |---|---|---| strategy | ~38 days | Successful patterns | fact | ~24 days | Preferences, identity | assumption | ~19 days | Inferred context | failure | ~11 days | Errors, environment-specific issues | Memory strength decays exponentially — but importance and recall frequency slow that decay: effective_λ = base_λ × (1 - importance × 0.8) strength = importance × e^(−effective_λ × days) × (1 + recall_count × 0.2) score = cosine_similarity × strength Memories recalled frequently resist decay. Memories below strength 0.05 are pruned automatically every 24 hours. Retrieval runs in two rounds to surface related context that vocabulary-based search misses: Round 1 — Vector search: cosine similarity against all memories, returns top-k above threshold. Round 2 — Graph expansion: BFS traversal from Round 1 seeds surfaces memories that share context but not vocabulary — connected via semantic edges (cosine similarity ≥ 0.4). recall("Python backend") Round 1 → [1] Python/MongoDB (sim=0.61) [2] DuckDB/spaCy (sim=0.19) Round 2 → [5] Docker/Kubernetes (sim=0.29 — below cut-off, surfaced via graph) Chain-aware pruning: A decayed memory is kept alive if any graph neighbour is above the prune threshold. Related memories age together. Multiple agents can share the same YourMemory instance — each with isolated private memories and controlled access to shared context. from src.services.api_keys import register_agent result = register_agent( agent_id="coding-agent", user_id="sachit", can_read=["shared", "private"], can_write=["shared", "private"], ) # → result["api_key"] — ym_xxxx, shown once only Pass api_key to any MCP call to authenticate as an agent: store_memory(content="Staging uses self-signed cert — skip SSL verify", importance=0.7, category="failure", api_key="ym_xxxx", visibility="private") recall_memory(query="staging SSL", api_key="ym_xxxx") # → returns shared memories + this agent's private memories # → other agents see shared only | Component | Role | |---|---| | DuckDB | Default vector DB — zero setup, native cosine similarity | | NetworkX | Default graph backend — persists at ~/.yourmemory/graph.pkl | | sentence-transformers | Local embeddings (all-mpnet-base-v2 , 768 dims) | | spaCy | Local NLP for deduplication and SVO triple extraction | | APScheduler | Automatic 24h decay job | | PostgreSQL + pgvector | Optional — for teams or large datasets | | Neo4j | Optional graph backend — pip install 'yourmemory[neo4j]' | PostgreSQL setup (optional) pip install yourmemory[postgres] Create a .env file: DATABASE_URL=postgresql://YOUR_USER@localhost:5432/yourmemory macOS brew install postgresql@16 pgvector && brew services start postgresql@16 createdb yourmemory Ubuntu / Debian sudo apt install postgresql postgresql-contrib postgresql-16-pgvector createdb yourmemory Claude / Cline / Cursor / Any MCP client │ ├── recall_memory(query, api_key?) │ └── embed → vector similarity (Round 1) │ → graph BFS expansion (Round 2) │ → score = sim × strength → top-k │ → recall propagation → boost neighbours │ ├── store_memory(content, importance, category?, visibility?, api_key?) │ └── question? → reject │ contradiction check → update if conflict │ embed() → INSERT → index_memory() → graph node + edges │ └── update_memory(id, new_content, importance) └── embed(new_content) → UPDATE → refresh graph node Vector DB (Round 1) Graph DB (Round 2) DuckDB (default) NetworkX (default) memories.duckdb graph.pkl ├── embedding FLOAT[768] ├── nodes: memory_id, strength ├── importance FLOAT └── edges: sim × verb_weight ≥ 0.4 ├── recall_count INTEGER ├── visibility VARCHAR Neo4j (opt-in) └── agent_id VARCHAR └── bolt://localhost:7687 Benchmarks use the LoCoMo dataset by Snap Research. Maharana et al. (2024). LoCoMo: Long Context Multimodal Benchmark for Dialogue. Snap Research. Copyright 2026 Sachit Misra — Licensed under CC-BY-NC-4.0. Free for: personal use, education, academic research, open-source projects. Not permitted: commercial use without a separate written agreement. Commercial licensing: [email protected]