Agent-Memory vs a-mem-mcp

Persistent memory for coding agents compared

Agent-Memory and a-mem-mcp both help coding agents retain and retrieve project knowledge across sessions. Agent-Memory focuses on capturing activity across supported agents, while a-mem-mcp builds connected memories from notes using an LLM.

Agent-Memorya-mem-mcp
Language—Python
LicenseApache-2.0MIT
Stars935
Forks43
Last analyzedNov 1, 2026Oct 28, 2026

Key differences

  • Agent-Memory captures agent activity through supported lifecycle hooks; a-mem-mcp adds notes through MCP tools or its Python API.
  • Agent-Memory combines keyword, vector, and knowledge-graph search; a-mem-mcp combines semantic search with graph traversal and evolves links between related notes.
  • Agent-Memory is a self-hosted server built on the iii engine, available as a native binary or through Docker; a-mem-mcp is a Python library and MCP server using ChromaDB.
  • Agent-Memory is licensed under Apache-2.0; a-mem-mcp is licensed under MIT.
  • Agent-Memory lists 6 stars and 1 fork; a-mem-mcp lists 35 stars and 3 forks. Its README identifies Claude Code as tested, with support for other MCP-compatible agents planned.

Choose Agent-Memory if you…

  • need activity capture and recall across supported coding agents.
  • want a shared memory server with MCP, REST API, and a local viewer.
  • prefer a setup available as a native binary or through Docker.
Read the Agent-Memory analysis →

Choose a-mem-mcp if you…

  • want connected notes that can be added and updated through MCP or a Python API.
  • need memory retrieval that combines semantic search with graph traversal.
  • use Claude Code or want configurable LLM backends, including Ollama.
Read the a-mem-mcp analysis →

Compare the numbers side by side, or add more projects

This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.