add_memory
Stores a piece of text or structural messages as a memory in Mem0.
LABSCO SUMMARY
Memory stops being one undifferentiated pile.
Each call carries a scope — user, agent, app, run or session — so a single store can serve several assistants without their recollections bleeding into each other. get_memory_capabilities reports what the configured backend actually supports, so the calls that exist only against the cloud service announce themselves instead of failing mid-task.
What it is
A client for Mem0: 16 tools that write memories from text or messages, search them with a similarity threshold and topK, and manage the records afterwards — updates, deletions, ratings, event logs and exports.
What you get
- Memories stored from a block of text or from structured messages, with metadata and custom instructions, and either waited on or left to complete in the background
- Search by query with a similarity threshold, topK, filters, a reference date and optional reranking
- Every read and write scoped by userId, agentId, appId, runId or sessionId
- Listing with pagination, single records fetched by ID, updated and deleted
- Bulk text updates and bulk deletions across many memory IDs in one call
- A memory's audit trail and revision log
- Quality feedback recorded against a memory record with a reason
- Background processing events listed and fetched individually by event ID
- Export jobs started with a schema and instructions, then polled for status and download
- The features the configured storage-mode backend supports reported back on request
Requirements
A Mem0 API key in MEM0_API_KEY and npx on your PATH. DEFAULT_USER_ID, DEFAULT_AGENT_ID and DEFAULT_APP_ID set the scope calls fall back to.
Setup effort
One command plus a key — npm install -g @pinkpixel/mem0-mcp, then supply credentials
USE CASES
01
Giving an assistant recall that survives the session
Write what matters as a memory — plain text or the message exchange it came from — with metadata attached, then search it back by meaning with a similarity threshold so the retrieval is not exact-match.
02
Keeping two assistants' memories apart
Every call takes userId, agentId, appId, runId or sessionId, so one Mem0 account can back several assistants and listing is scoped to the one you are asking about.
03
Correcting something the assistant now believes wrongly
Fetch the record by ID to see what it actually says, update the text if it needs rewording, and delete it when it should not have been remembered at all.
04
Clearing out a run's memories in bulk
List what is held under that run or session, then delete the IDs in a single confirmed call instead of one request per record.
05
Auditing how a memory got there
The revision log shows what a record used to say, and the background event log shows the processing jobs that wrote it — which is what you need when a memory looks wrong and nobody typed it.
06
Taking the store somewhere else
An export runs as a job against a schema and instructions you set, then is polled for status and download, so moving or archiving the memory set is not a scrape.
16 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
batch_delete_memories
Performs bulk deletions of multiple memories.
batch_update_memories
Performs bulk updates of text contents for multiple memories (cloud only).
create_memory_export
Kicks off an async memory export query job (cloud only).
delete_memory
Deletes a specific memory record by ID.
get_memory
Gets a single memory by its ID.
get_memory_capabilities
Returns the features and APIs supported by the current storage mode backend.
get_memory_event
Manually retrieves detail logs of a specific background event job (cloud only).
get_memory_export
Retrieves status and downloads of a memory export job (cloud only).
get_memory_history
Gets the audit trail and revision log of a memory (cloud only).
list_memories
Lists all memories scoped to specific identifiers with pagination.
list_memory_events
Lists history logs of background memory processing events (cloud only).
rate_memory
Submits quality feedback evaluation for a memory record (cloud only).
search_memories
Searches stored memories based on a query.
search_memory
Backward-compatible alias for search_memories.
update_memory
Updates a specific memory record.
BEFORE YOU INSTALL
What to know first
Tools you get
16 tools when we connected. Seven tools are cloud-only, so Supabase or local storage leaves nine usable. search_memory is an alias for search_memories.
Requires
- npx on PATH
- MEM0_API_KEY in your environment
Optional
- SUPABASE_URL in your environment
- SUPABASE_KEY in your environment
- OPENAI_API_KEY in your environment
CONTEXT COST
Light Standing cost
Rated by the size of the 16 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
TERMINAL
npm install -g @pinkpixel/mem0-mcpMCP Clients
Claude Desktop, Cursor, Cline, Roo Code
JSON
{
"mcpServers": {
"mem0-mcp": {
"command": "npx",
"args": [
"-y",
"@pinkpixel/mem0-mcp"
],
"env": {
"MEM0_API_KEY": "<YOUR_MEM0_API_KEY>",
"DEFAULT_USER_ID": "user123",
"DEFAULT_AGENT_ID": "<YOUR_DEFAULT_AGENT_ID>",
"DEFAULT_APP_ID": "<YOUR_DEFAULT_APP_ID>"
}
}
}
}The configuration on this page does not set SUPABASE_URL, SUPABASE_KEY and OPENAI_API_KEY; the project's instructions put them in a .env file. To supply them without depending on where the server looks for that file, use the block's "env" section.
Claude Desktop — Smithery
CLAUDE DESKTOP — SMITHERY · TERMINAL
npx -y @smithery/cli install @pinkpixel-dev/mem0-mcp-server --client claude
