AgentmemoryInfinite memory for your AI agents at a fraction of the token cost

Agentmemory

Agentmemory is the memory layer your coding agent should have had from day one. It captures every session, recalls memories in milliseconds, and runs anywhere as a single Node process with zero external databases.

Product Highlights

• Zero External Dependencies: No Redis, Kafka, Postgres, or Qdrant. The entire runtime is one self-contained process.
• Triple-Stream Retrieval: BM25 + Vector + Knowledge Graph hybrid search achieves 95.2% R@5 on LongMemEval-S, with P50 under 20ms on a laptop.
• 12 Auto-Capture Hooks: Every tool call, prompt, and stop becomes a compressed observation without glue code.
• Native MCP Surface: 51 MCP tools including memory_save, memory_recall, and governance functions.
• Auto Consolidation: Hourly sweeps compress raw observations into semantic memories, merge duplicates, and decay stale rows.

Use Cases

• AI Coding Assistants: Give Claude Code, Cursor, or VS Code agents persistent memory across sessions.
• Session Replay & Audit: Rehydrate full sessions from JSONL transcripts with complete timeline reconstruction.
• Knowledge Graph Building: Extract entities and relations automatically, query via REST, and visualize connections.
• Peer-to-Peer Memory Sync: Federate memories across multiple agentmemory nodes with authenticated HTTPS.

Target Audience

Agentmemory is built for developers and teams running AI coding agents who need fast, private, persistent memory without managing complex database infrastructure. Ideal for those prioritizing data locality and low-latency retrieval.