a-mem-mcp vs deja-vu
Coding-agent memory and session-history search compared
a-mem-mcp builds a persistent graph of notes that agents can add and connect over time. deja-vu indexes existing coding-agent sessions for local search, so the main difference is created memory versus searchable conversation history.
a-mem-mcp: Build Evolving Memory for Coding Agents
A-MEM is an MCP server and Python library that stores agent knowledge as an evolving, connected memory graph. It suits coding-agent users who need to retrieve and build on project context across sessions.
deja-vu: Search Coding-Agent History for Reusable Memory
deja-vu is a Go-based local memory tool that indexes existing session history from Claude Code, Codex, Cursor and 39 more coding agents. It provides search, MCP recall and hooks without requiring an LLM, so agents can reuse earlier decisions and fixes.
| a-mem-mcp | deja-vu | |
|---|---|---|
| Language | Python | Go |
| License | MIT | MIT |
| Stars | 35 | 1.2k |
| Forks | 3 | 128 |
| Last analyzed | Oct 3, 2026 | Oct 19, 2026 |
Key differences
- a-mem-mcp stores LLM-enriched notes and links related memories; deja-vu indexes existing session histories and work details such as commands and edit spans.
- a-mem-mcp uses semantic search with graph traversal; deja-vu supports search without embeddings, with semantic search available through a configured embedding endpoint.
- a-mem-mcp is a Python library and MCP server; deja-vu provides a command-line interface and an MCP server, and is written in Go.
- a-mem-mcp documents ChromaDB persistence and configurable LLM backends; deja-vu indexes session files locally and does not require an LLM.
- a-mem-mcp reports 35 stars and 3 forks; deja-vu reports 1.1k stars and 109 forks, and 28 open issues.
- a-mem-mcp reports testing with Claude Code and describes broader MCP-agent support as planned; deja-vu indexes histories from agents including Claude Code, Codex, and Cursor, with integrations varying by harness.
Choose a-mem-mcp if you…
- need agents to save project decisions and build connected memory across sessions.
- want an MCP server plus a Python API for integrating persistent memory.
- need configurable LLM backends, including local Ollama, for memory processing.
Choose deja-vu if you…
- want to search existing coding-agent conversations rather than save selected notes into a new memory system.
- switch between coding agents or need to recover past commands, fixes, and file discussions.
- need local search that works without an LLM or embedding service.
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.