Keymem MCP Server

The associative memory layer for LLM agents - recall by association, not just similarity. It runs locally over stdio via the published package.

People who work with Openai and Embedding and want it reachable from Claude, Cursor, VS Code, or another MCP client. The project is written in TypeScript.

VERIFIED ACTIVE

LAST COMMIT 2026-09-17 · ★ 3 · #66 OF 304 MAINTAINED MEMORY · VERIFIED 2026-09-18

MIT · TypeScript servers · how we verify → /methodology

01 · Install Keymem

before you install - you'll need

Set OPENAI_API_KEY, EMBEDDING_BACKEND, OPENAI_EMBEDDING_MODEL, KEYMEM_KEY_RECALL, KEYMEM_MEMORY_DEDUP, KEYMEM_RERANK_MIN_SCORE before connecting. LOCAL_EMBEDDING_MODEL is optional or environment-specific per the README.

Claude Code

claude mcp add donggyun112-keymem -- npx -y keymem

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "donggyun112-keymem": {
      "command": "npx",
      "args": [
        "-y",
        "keymem"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license MIT - declared in the repository

npm package keymem - unscoped; check the name against the project README before installing

registry namespace io.github.donggyun112 is GitHub-verified and matches the repo owner

03 · What Keymem can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

Limitations (from the README)

- Linear scan - suitable for personal use (~10k memories). FAISS/ChromaDB integration planned for larger scale. - Later-hop round trips - recall completes the first Key → Memory hop, but following a connected key to another memory still requires read_key → read_memory. - Hub breadth - broad keys can connect many memories. read_key() paginates hubs; the agent must choose whether to continue paging or follow a more spe

Latest releases

v0.29.0 · 2026-09-12

Added · Task-conditioned evidence selection for comparison-shaped expanded recall and injection. Broad · graph activation is preserved; a conservative named-entity controller projects the activated pool · onto the…

v0.24.0 · 2026-08-30

Added · dismiss(memory_id, key_id) - the graph's first negative signal. Every signal it carried · was positive: a read deepens a memory and strengthens the edge it traversed, a weak match · accrues heat toward becoming…

v0.23.0 · 2026-08-30

Fixed · Namespaces are case-folded on write, on query, and on load. They were compared · with exact string equality everywhere, so Nexora and nexora hard-partitioned the · same project and a scoped recall silently…

04 · Who maintains Keymem

keymem is maintained by donggyun112. We track 2 MCP servers from donggyun112 - 2 actively maintained, 5 combined GitHub stars, oldest repo from Mar 2026. Full record: all servers from donggyun112.

codecanvas Python code intelligence for coding agents: call paths, control flow, and change impact. ★ 2

05 · Facts

repository: github.com/donggyun112/keymem
category: memory - ranked #66 of 304 actively-maintained memory servers as of 2026-09-18.
release cadence: 8 releases in the last 90 days (latest 2026-09-12)
registry: io.github.donggyun112/keymem (active, first published 2026-06-22 · 15 versions)
packages: npm:keymem

06 · Keymem FAQ

What is Keymem?

The associative memory layer for LLM agents - recall by association, not just similarity. It runs locally over stdio via the published package.

Is Keymem still maintained?

Yes - as of 2026-09-18, its last commit was 2026-09-17 and it shipped 8 releases in the last 90 days. We re-verify nightly.

How do I install Keymem?

Run `npx -y keymem`. The README documents 7 environment variables (OPENAI_API_KEY, EMBEDDING_BACKEND, LOCAL_EMBEDDING_MODEL…) to set first. Set OPENAI_API_KEY, EMBEDDING_BACKEND, OPENAI_EMBEDDING_MODEL, KEYMEM_KEY_RECALL, KEYMEM_MEMORY_DEDUP, KEYMEM_RERANK_MIN_SCORE before connecting. LOCAL_EMBEDDING_MODEL is optional or environment-specific per the README. You can also paste the ready-made client config above.

Does Keymem run locally?

Yes - it's a stdio server: it runs on your machine (via npx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to Keymem

Alternatives to Keymem

Maintained memory servers if Keymem isn't the fit.

Pairs well with

Servers that cover what Keymem doesn't - only shown when the pairing reason fits the companion.

More memory MCP servers · Mitos · Openeye · Jprofiler · Emercoin Agent · Dbgprobe MCP Server

More TypeScript MCP servers · Domain Search MCP · Hivelore · Megalinter · Arena · Help Scout MCP Server · see all