Recall MCP
An MCP server that keeps project decisions in Redis or Valkey and brings them back in new sessions and after compaction.
- 178 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Memory captures everything discussed in conversations, including secrets and customer data.
- Securing Redis or Valkey (password, TLS, network) remains your job.
Install
- Works in
Manual install
npx @modelcontextprotocol/create-server @joseairosa/recallThe first option from the README installation section. A MCP config with `npx -y @joseairosa/recall` follows it.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Memory captures everything discussed in conversations, including secrets and customer data.
- Securing Redis or Valkey (password, TLS, network) remains your job.
This is third-party code. Review the repository files before installing.
What it does
Recall is a persistent-memory MCP server for Claude and other agents. Decisions, patterns and knowledge are stored once and the agent finds them in later conversations by itself. The server exposes 27 tools: `store_memory`, `search_memories`, `recall_relevant_context`, `summarize_session`, `export_memories`, `find_duplicates`, `link_memories` and more. Storage is your own Redis or Valkey. Search is semantic: connect one of seven embedding providers or fall back to keywords. Memories split into workspace and global ones, can be linked, versioned and merged when similar. A Claude Code plugin adds commands, agents and hooks. For those who do not want their own Redis, the author runs a paid cloud service at recallmcp.com.
Who it is for. Developers tired of repeating architectural decisions to an agent who want project memory on their own server.
Good fit when
- An agent loses agreements after a session restart or context compaction.
- Project decisions must move between machines through a shared Redis.
- A team needs shared memory split by workspace.
- Duplicate notes piled up and must be found and merged.
Not a fit when
- Redis or Valkey is unavailable and you do not want to run one.
- Memory would hold secrets that must not be stored even on your own server.
- A project instruction file such as `CLAUDE.md` is enough.
Example request
Store the decision to pick PostgreSQL over MongoDB with reasons, then in a new session find all database decisions.MCP
- Transport
- stdio
- Authentication
- not required
Environment variables
`REDIS_URL`secret- Redis or Valkey address, `redis://localhost:6379` by default.
Limitations
Redis 6 or newer or Valkey is required, plus Node.js 18 or newer. Memory holds everything discussed in conversations: code snippets, keys, architecture, people's names. Securing the store (password, TLS, network limits) is your job and the README says so in a dedicated warning. The author accepts no liability for leaks or data loss. Some features, including the RLM plugin, are described together with the cloud service, so check what works in a self-hosted setup.
How to disable. Remove the `recall` entry from the MCP configuration and the plugin from `~/.claude/plugins/recall-rlm/`. Clean data in Redis or Valkey separately.
FAQ
Is the cloud service required?
No. The repository runs on your own Redis or Valkey. The recallmcp.com cloud is a separate paid option with an API key.
Which tools exist?
Twenty-seven: store and update, search, session organization, export and import, duplicate detection, links between memories, global and workspace scopes.
How do I connect it?
Add the server to the MCP configuration with `npx -y @joseairosa/recall` and set `REDIS_URL`. A global install via `npm install -g @joseairosa/recall` also exists.
Can memory be shared between machines?
Yes, by pointing to the same `REDIS_URL` on a cloud Redis or Valkey.
What does the Claude Code plugin do?
Adds commands, agents and hooks for large files and RLM features. It installs through a script from the README.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server that preserves Claude Code context between sessions: decisions, checkpoints and a file cache kept in local SQLite.
MemoMind MCP gives Claude Code project memory, with stored facts, events and conclusions visible in a local dashboard.
Local memory for agents: stores conversations and project files verbatim and retrieves them with semantic search over MCP
Agent memory as a graph of linked facts: decisions and errors outlive the session, with no vector database or paid API calls
A long-term memory MCP server for agents: one Rust binary that runs offline and never calls a language model
Self-hosted long-term memory for agents: semantic search, a knowledge graph and autonomous consolidation over MCP and a REST API
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Recall (this card) | no | not required | 178 | |
| agentmemory | yes | not required | 30,160 | another agent memory implementation; Recall stores data in Redis or Valkey and tracks links and versions of memories |
| MCP Memory Service | no | not required | 2,090 | a separate memory service with its own storage; Recall relies on your Redis and 27 tools |
README in short
Memory lives in your Redis or Valkey, search is semantic through the chosen embedding provider and a security warning opens the README.