ai-memory-mcp
ai-memory-mcp is an MCP server, listed under databases. It reaches sqlite and xai. It needs AI_MEMORY_LLM_API_KEY and XAI_API_KEY to authenticate. Its repository has 54 GitHub stars and is licensed under Apache-2.0; we last read it on 2026-10-22.
Stores persistent memory for AI models including Claude, ChatGPT, Grok, Gemini, Codex, Cursor, OpenClaw, and MCP clients. Provides an HTTP API and CLI interface. Uses SQLite FTS5 and has no cloud dependencies.
Install
Read from ai-memory-mcp’s own README, close to verbatim -- not written by us, and not guessed.
npm install -g @alphaone/ai-memory
What it reaches, and what it needs
What it connects to
Unverified means a mention in the server’s own description, not a citation. A wrong claim here costs nothing to correct and nothing to trust yet.
Maintenance
Persistent memory for any AI — MCP server, HTTP API, CLI. Works with Claude, ChatGPT, Grok, Gemini, Codex, Cursor, OpenClaw, and any MCP client. One Rust binary; SQLite or PostgreSQL + Apache AGE + pgvector; rust-a2a wake plane. Apache-2.0.
What ai-memory-mcp can do
3 tools documented in its README. An AI client calls these by name.
invocation_record- Tracks the invocation of a skill, such as B7-SKILL, with details like the invocation ID and timestamp.
recall_observations- Enables a shadow-feedback loop for recalling observations, as seen in issue [#1706](https://github.com/alphaonedev/ai-memory-mcp/issues/1706).
memory_lineage- Provides a memory-derivation lineage DAG, as seen in issue [#1859](https://github.com/alphaonedev/ai-memory-mcp/issues/1859).
Read from the README; names are kept only if the README contains them.
ai-memory-mcp compared with similar servers
| Server | Reaches | Needs | Stars | Last update | Licence |
|---|---|---|---|---|---|
| ai-memory-mcp (this page) | sqlite, xai | AI_MEMORY_LLM_API_KEY, XAI_API_KEY | 54 | 2026-10-22 | Apache-2.0 |
| dbhub | mysql, postgres, sqlite(1 shared) | None listed | 3,702 | 2026-10-21 | MIT |
| mcp-alchemy | mysql, postgresql, sqlite(1 shared) | None listed | 422 | 2026-10-15 | MPL-2.0 |
| knowl | sqlite(1 shared) | None listed | 114 | 2026-10-21 | Apache-2.0 |
| saga-mcp | sqlite(1 shared) | None listed | 40 | 2026-10-07 | MIT |
| remem | project, sqlite(1 shared) | PROJECT_KEY | 34 | 2026-10-21 | MIT |
Ranked by services in common with ai-memory-mcp, then by stars. Figures are from each repository when we last read it.
Common questions about ai-memory-mcp
- What does ai-memory-mcp do?
- Stores persistent memory for AI models including Claude, ChatGPT, Grok, Gemini, Codex, Cursor, OpenClaw, and MCP clients. Provides an HTTP API and CLI interface. Uses SQLite FTS5 and has no cloud dependencies.
- How do I install ai-memory-mcp?
- Its own README gives this command: npm install -g @alphaone/ai-memory. Run it where your MCP client can launch the server, and check the README for any settings it asks for.
- What does ai-memory-mcp connect to, and what does it need?
- It reaches sqlite and xai. The credentials we found in its source are AI_MEMORY_LLM_API_KEY and XAI_API_KEY.
- Which AI clients can use ai-memory-mcp?
- Its own description mentions Cursor. That is a mention in its description, not a test we ran.
- Is ai-memory-mcp official and is it maintained?
- We found nothing naming it as a vendor’s official server. Its repository has 54 GitHub stars and its state in our directory is listed. Its last commit we recorded was on 2026-10-22.
- What licence does ai-memory-mcp use?
- The repository declares the Apache-2.0 licence.
More MCP servers for databases
- CodeGraphContext4,360 stars
- dbhub3,702 stars
- mcp-server-mysql2,173 stars
- mysql_mcp_server1,402 stars
- server874 stars
- supabase-mcp-server833 stars