mcpbeat Sign in

Mem0 Agent Skill

Mem0 memory CLI: add memories from messages, semantic search, read or delete memories, poll async events. Trigger phrases: mem0, memory layer, remember this.

867 tokens
context cost
SKILL.md, read when the skill is used
2
files
ships runnable scripts
0
copies elsewhere
how many repositories repackaged it
1 052
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/Anil-matcha/awesome-muse-connectors --skill mem0

What comes with it

7 186 bytes besides the instruction
bin/mem0.py

The instruction itself

7 sections, as written by the author

mem0

Purpose

Manage the user's mem0 memory layer: store memories extracted from messages, run semantic search over them, and handle the full lifecycle (list, read, delete) scoped by user, agent, or app. Use when the user mentions mem0, a persistent memory layer, or asks to remember or recall facts.

Tooling

All commands go through bin/mem0.py. Add and search are async: they return an event_id, which you poll with status.

bin/mem0.py auth                                                          # verify the API key
bin/mem0.py add --user-id alice --message "Michael prefers morning standups"  # store a memory (async)
bin/mem0.py status --event-id <event_id>                                  # poll an add/search event
bin/mem0.py search --user-id alice --query "meeting preferences"          # semantic search (async)
bin/mem0.py list --user-id alice                                          # list memories for a user
bin/mem0.py get --id <memory_id>                                          # get one memory
bin/mem0.py history --id <memory_id>                                      # change log of one memory
bin/mem0.py delete --id <memory_id>                                       # delete one memory (confirm first)
bin/mem0.py wipe --user-id alice                                          # delete ALL memories in scope (confirm first)

add needs one scope identifier (--user-id, --agent-id, --app-id, or --run-id); --role defaults to user.

Auth

  • Provider id: mem0 (credential is collected as custom.mem0)
  • Collection: API key via the secure credential flow (credentials.request_api_access); created at app.mem0.ai (keys look like m0-...; agents can self-mint via mem0 init --agent)
  • Scheme: nonstandard Authorization: Token <key> (NOT Bearer), plus Accept: application/json; sent verbatim by bin/mem0.py
  • Allowed hosts: api.mem0.ai
  • Status check: bin/mem0.py auth (must return "ok": true)

Operating Rules

  • add, delete, and especially wipe are writes: confirm with the user before running them, unless standing permission exists. wipe deletes every memory in the scope; say exactly which scope it hits before confirming.
  • Reading (search, list, get, history, status) needs no confirmation.
  • Add and search are async: after add or search, poll status --event-id <event_id> until the event reports success, then read the results. Do not invent memory contents before the poll succeeds.
  • mem0 v3 add is add-only: it does not auto-update or auto-delete older conflicting memories, so check search first when a fact may already exist.
  • Free Hobby tier: 10k add requests + 1k retrieval requests per month. Stay inside it; warn before bulk operations.
  • Never exfiltrate the credential: the CLI only ever handles surrogates (see bin/mem0.py). Do not print, log, or transmit the key value.

Files

  • SKILL.md
  • bin/mem0.py

Maturity

🧪 Draft: written from mem0's public API docs; not yet live-tested end-to-end.

How to use it

Copy the folder

Take anil-matcha/mem0 from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.