Mnemo
Persistent memory for Claude Code. Your AI never starts from scratch again.

Agenten stärken
Installation vom Herausgeber. Kein PlugMCP-Endpunkt.
Host a copy with our keys and logs.
Noch kein paketiertes Release in der Registry. Folgen Sie der README des Repositorys.
Remote-URL oder ein Befehl auf Ihrem Rechner? Worin sich die beiden Installationen unterscheiden
Auf PlugMCP ausführen
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
StartenMnemo gives Claude Code a brain that survives across sessions. It captures decisions, conventions, and preferences — and recalls them by meaning, on demand, with sub-100ms semantic search.
# install the CLI
npm install -g @mnemo-mcp/cli
# install Mnemo into Claude Code (skill + MCP server)
mnemo init
# (optional) also wire auto-capture hooks
mnemo init --with-hooks
# verify
mnemo doctor
After mnemo init, restart Claude Code. The /mnemo skill and the mnemo_* MCP tools are now available — Claude will use them automatically when relevant.
mnemo remember "our API auth uses OAuth2 with refresh tokens every 30min"
mnemo remember --global "I prefer pnpm over npm"
mnemo recall "what's our auth pattern?"
mnemo recall "auth" --explain # show the ranking breakdown per hit
mnemo list
mnemo stats
mnemo forget <id>
mnemo backup # timestamped snapshot
mnemo restore mnemo-backups/<file>.json
mnemo migrate # rebuild the index after an embedder change
mnemo serve # localhost web dashboard
mnemo watch docs/ # auto-capture matching files as they change
mnemo entity create AuthService -t service -d "login + tokens"
mnemo entity link Billing requires AuthService
mnemo entity attach <memoryId> AuthService
mnemo entity show AuthService --depends # memories, relations, and what depends on it
mnemo agents # who captured what (claude-code, cursor, aider…)
mnemo plugins # discovered embedder/ranker/hook plugins
# end-to-end encrypted multi-machine sync (server stores ciphertext only):
export MNEMO_ENCRYPTION_KEY=… MNEMO_SYNC_URL=https://your-host MNEMO_SYNC_TOKEN=…
mnemo sync push # encrypt locally → upload
mnemo sync pull # download → decrypt → import
Run your own sync backend with npx @mnemo-mcp/sync-server — it never sees plaintext.
export MNEMO_ENCRYPTION_KEY="a strong passphrase" # memory.db is then AES-256-GCM at rest
Claude Code forgets everything when the session ends. CLAUDE.md partially helps — it's a static blob loaded on every turn — but it grows unwieldy fast and can't answer "do we have a precedent for X?"
Mnemo gives you semantic, on-demand memory:
Claude Code session
├── /mnemo skill ← teaches Claude when to call the tools
├── @mnemo-mcp/server server ← exposes recall/remember/forget/list/stats
└── @mnemo-mcp/cli hooks ← session-start, pre-task, post-edit auto-wiring
│
▼
@mnemo-mcp/core
├── ONNX all-MiniLM-L6-v2 (384-dim embeddings, ~25MB, lazy)
├── HNSW vector index (sub-100ms recall at 50k memories)
└── sql.js (WASM SQLite) (~/.mnemo/memory.db)
Everything is local-first. No daemon. No cloud account. No telemetry.
By default the post-edit hook captures any change to:
CLAUDE.md, AGENTS.md, GEMINI.md*.adr.mddocs/decisions/**docs/adr/**Anything you write to those files becomes a project memory automatically. Disable with mnemo init (no --with-hooks).
score = 0.7 × cosine_similarity + 0.2 × recency + 0.1 × access_boost
Recency decays with a 30-day half-life. Access boost saturates at 20 reads.
@mnemo-mcp/core — pure TS memory engine (embeddings, HNSW, store, graph, plugins, crypto)
@mnemo-mcp/server — MCP server (drop into Claude Code; 9 tools)
@mnemo-mcp/cli — terminal commands + hook handlers + init
@mnemo-mcp/sync-server — self-hostable, end-to-end-encrypted sync backend
/mnemo skillmnemo init, beefier doctor, CI, polishmnemo procedure record/run/suggest/done + 2 new MCP tools + auto-suggestion via pre-task hookbackup/restore, migrate, watch, serve dashboard, recall --explain/--stream, encryption at rest, int8 quantizationStop/SubagentStop hooks)mnemo agents, standard MCP discovery)@mnemo-mcp/sync-server + mnemo sync) + web UISee ROADMAP.md for design notes on each.
npm install
npm run build
npm test # fast (150+ tests, hash embedder)
MNEMO_TEST_ONNX=1 MNEMO_E2E=1 npm test # full incl. real ONNX (~5s)
npm run lint
MIT — see LICENSE.
Persistent memory for Claude Code. Your AI never starts from scratch again.
Tools
Noch keine Tool-Liste. Remote-Server erhalten eine nach dem ersten Health-Check.
FAQ
Kopieren Sie auf dieser Seite die stdio-Konfiguration für Cursor, Claude Desktop, Claude Code oder VS Code. Sie nutzt die Installation des Herausgebers, keinen PlugMCP-Proxy.
In diesem Eintrag ist keine erforderliche Umgebungsvariable aufgeführt.
Mnemo läuft lokal über stdio.
Zuletzt geprüft: 2026-10-28. Persistent memory for Claude Code. Your AI never starts from scratch again.