agentmemory
#1 Persistent memory for AI coding agents based on real-world benchmarks
What is agentmemory?
#1 Persistent memory for AI coding agents based on real-world benchmarks
Overview of agentmemory
What it is
AgentMemory is a persistent memory system for AI coding agents. It captures coding sessions, makes decisions and fixes searchable, and lets agents recall them later. It extends Karpathy's LLM Wiki pattern with confidence scoring, lifecycle, knowledge graphs, and hybrid search. It works with Claude Code, GitHub Copilot CLI, Cursor, Gemini CLI, Codex CLI, Hermes, OpenClaw, pi, Oh My Pi, OpenCode, DeepSeek Harness, and any MCP client.
Key features
- Capture: native hooks and plugins record supported session events, with secret scrubbing and offline replay.
- Recall: keyword search works without an API key; local or remote embeddings add semantic search.
- Continue: retrieve project history, lessons and session handoffs from another agent.
- Inspect: browse, edit, export and forget memories in the live viewer.
- Keyless mode uses BM25 for recall; on-device semantic recall available with
EMBEDDING_PROVIDER=local. - Supports multiple agents via hooks, MCP, or REST API; all agents share the same memory server.
Use cases
- Persistent memory across coding sessions for AI coding agents.
- Sharing project history across different tools and agents.
- Retrieving scoped memories and inspecting original sources in a local viewer.
- Offline replay and secret scrubbing for session events.
Getting started
Requirements: Node.js 20 or newer with npm and npx. Start setup: bash npx -y @agentmemory/agentmemory@latest Then seed sample sessions and add skills: bash npx -y @agentmemory/agentmemory@latest demo npx skills add rohitg00/agentmemory -y The API is at http://localhost:3111 and the live viewer at http://localhost:3113.
License and pricing
Licensed under Apache-2.0. Open source.
Based on the project's public README and repository metadata.