Provides tools for reading upcoming Google Calendar meetings, resolving attendees to known people, and generating pre-meeting briefs with citations and previous meeting context.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@LedgerMem MCP Server Remember that my name is Alice and I work at Acme Corp."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Model Context Protocol server for Mnemo Memory — exposes long-term memory tools to any MCP client (Claude Desktop, Cursor, Windsurf, VS Code, Zed).
Tool | What it does |
| Hybrid 7-strategy retrieval over the workspace memory store. Filter by |
| Cited-answer pipeline: question in, synthesized answer + citations out ( |
| Store an atomic fact with optional metadata; |
| Fetch one memory by ID within the configured container. |
| Patch an existing memory's content, metadata, or polarity. |
| Soft-delete a memory by ID so it leaves retrieval. |
| Undo a soft delete during its recovery window (API-key sessions only). |
| Paginate through memories (cursor-based). |
| Ingest a raw document (up to 500,000 characters) into the async extraction lane; returns a |
| Poll an ingestion job: queued / processing / completed / failed (API-key sessions only). |
Tool | What it does | Key scope(s) | API flag |
| Today's reminders, important dates, last-24h captures, follow-ups and meetings for one container. |
|
|
| One container as a time-ordered stream of memories, reminders, documents (and optional events). |
|
|
| People the workspace remembers, with memory / open-reminder counts. |
|
|
| One person by slug (contact details, important dates, aliases). |
|
|
| Create a person, or update them when the slug already exists. |
|
|
| A reminder with a due time, filed under a person or a container. |
|
|
| Overdue / due-today / upcoming reminders plus important dates, by local day. |
|
|
| Mark a reminder done. |
|
|
| Upcoming Google Calendar meetings with attendees resolved to people. |
|
|
| Pre-meeting brief: reader answer with citations, matched people, previous meetings. |
|
|
| Fold 2..20 duplicate memories into one survivor (idempotent on |
|
|
These scopes are opt-in: mint (or edit) the API key with them at
https://app.mnemohq.com/settings/api-keys. Every feature is dark until the
API operator sets its MEMORY_API_* flag; until then the tool returns a
503 FEATURE_DISABLED error that names the flag.
All personal-memory tools except memory_timeline are API-key only:
the API rejects hosted OAuth MCP tokens on cross-container routes, so hosted
OAuth sessions (https://mcp.mnemohq.com/mcp) do not list them.
How the team does things, learned from their captured browser work. Read only, published procedures only, and the same for every caller.
Tool | What it does | Scope |
| The workspace's published procedures, most recently seen first, optionally for one app. |
|
| One procedure's current steps and conditions, plus its version history. |
|
| "How do we do X here?", answered from published procedures with citations. |
|
Hosted OAuth sessions get all three. A connection limited to some containers
reaches them only when its grant includes procedures:workspace.
Remote HTTP calls are scoped by the user's Mnemo OAuth grant. The user signs in, chooses a workspace and memory container, and the server derives that scope from the access token. The model cannot set or change it.
Related MCP server: Memorium
npx -y getmnemo-mcpOr wire it into the client config directly. Example for Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"getmnemo": {
"command": "npx",
"args": ["-y", "getmnemo-mcp"],
"env": {
"GETMNEMO_API_KEY": "prfly_live_...",
"GETMNEMO_WORKSPACE_ID": "ws_...",
"GETMNEMO_CONTAINER_TAG": "user:jane"
}
}
}
}Get an API key at https://app.mnemohq.com/settings/api-keys.
For clients that support remote MCP and OAuth 2.1, use:
https://mcp.mnemohq.com/mcpThe client should discover OAuth metadata, open Mnemo sign-in, and return to the client after the user selects a workspace and container. No API key, workspace header, or container configuration is required for this path.
npm install
cp .env.example .env # fill in API-key/container values for stdio
npm run dev # stdio
npm run dev:http # Streamable HTTP on :8787
npm run build # bundle to dist/
npm run lint && npm run typecheck && npm teststdio (src/cli.ts): one process per MCP client connection, env-configured.
Streamable HTTP (src/http.ts): hosted transport for public OAuth sessions, with private API-key compatibility for trusted deployments.
Both transports share src/server.ts (tool registration + dispatch), src/personal-tools.ts (personal-memory tools) and src/api-client.ts (typed REST wrapper).
The public hosted transport uses Mnemo's OAuth 2.1 authorization server and resource-bound tokens. Header-selected containers remain disabled by default; they are only a private compatibility path for a trusted gateway.
The server deliberately does NOT depend on getmnemo (the JS SDK) so it can ship independently.
MIT
This server cannot be deployed
Persistent personal memory for AI assistants — save, search, and recall across every MCP client.
Hosted agent memory: store, search, and recall facts across sessions from any MCP client.
Persistent memory for AI assistants. Save once; recall from Claude, ChatGPT, or any MCP client.
Persistent memory for AI agents to retain, retrieve, and recall conversation context through MCP.
