Skip to content
MCP server

RE-call MCP Memory Server

Agent memory in SQLite or Postgres that flags stale facts and abstains when unsure

Runs locallyStatus and history
Pricing
Open source
Platforms
CLI, self-hosted, Windows
Runs
Locally
Version
0.15.0
Listing updated
3 days ago

About RE-call MCP Memory Server

RE-call MCP Memory Server is an open-source MCP server for knowledge and memory. It runs locally on your machine and is started by your AI client.

Plain vector search returns the nearest note whether or not it answers the question. Here, results pass a calibrated trust gate: a claim that has been superseded is marked as such, and a query with no confident match returns ABSTAIN with a reason and the weak evidence, such as a match at confidence 0.30, instead of a guess.

Two editions share that trust layer. RE-call Lite keeps one project's memory in a local SQLite file and refreshes changed notes automatically, with no database server or Docker. The PostgreSQL edition adds pgvector, shared serving, authentication and immutable calibrated generations. Retrieval is hybrid dense plus full text, embeddings run locally by default, and no LLM call is needed in the memory layer.

It is a Python package (3.11 and up) that exposes an MCP server for Claude Code, Codex or any client, and also integrates with the Claude Agent SDK, LangChain and LlamaIndex. It is AGPL-3.0 with a commercial licence available.

Install

Run in a terminal

claude mcp add re-call --env RECALL_SERVING_DSN='<RECALL_SERVING_DSN>' -- uvx recall-rag

Add to claude_desktop_config.json

{
  "mcpServers": {
    "re-call": {
      "command": "uvx",
      "args": [
        "recall-rag"
      ],
      "env": {
        "RECALL_SERVING_DSN": "<RECALL_SERVING_DSN>"
      }
    }
  }
}

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "re-call": {
      "command": "uvx",
      "args": [
        "recall-rag"
      ],
      "env": {
        "RECALL_SERVING_DSN": "<RECALL_SERVING_DSN>"
      }
    }
  }
}

Add to .vscode/mcp.json

{
  "servers": {
    "re-call": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "recall-rag"
      ],
      "env": {
        "RECALL_SERVING_DSN": "<RECALL_SERVING_DSN>"
      }
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "re-call": {
      "command": "uvx",
      "args": [
        "recall-rag"
      ],
      "env": {
        "RECALL_SERVING_DSN": "<RECALL_SERVING_DSN>"
      }
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "re-call": {
      "command": "uvx",
      "args": [
        "recall-rag"
      ],
      "env": {
        "RECALL_SERVING_DSN": "<RECALL_SERVING_DSN>"
      }
    }
  }
}

Configuration

  • RECALL_SERVING_DSNThe store RE-call reads: a PostgreSQL DSN, or the sqlite:/// address `recall setup --lite` prints for a local Lite store with no database server.Required
  • RECALL_TENANTTenant id served by this local stdio server.
  • RECALL_TRUST_MODESet to development only for local uncalibrated corpora.
  • RECALL_EMBEDDERThe embedder the corpus was BUILT with. This registry command installs FastEmbed: use fastembed or a supported FastEmbed model such as BAAI/bge-large-en-v1.5. A Voyage corpus requires a separate launch installed with recall-rag[mcp,voyage] and its API key. The model must match the corpus even when dimensions agree.
  • RECALL_TABLEThe chunk table inside that database. `recall quickstart` indexes into quickstart_chunks so its sample corpus can never be retrieved beside real memory; `recall setup` uses chunks. The wrong value returns no hits and no error.
  • RECALL_MCP_TOOLSOptional. Serve a subset of the tools: a preset (all, search, read) or explicit names. Every tool definition is re-sent every turn, so serving ones nobody calls costs input tokens on every turn. Unset serves them all.

Tools this server provides (1)

Taken from the project documentation; the running server may differ.

recall_search
Searches memory through the trust gate and returns a verdict or an abstention.

What it does

Abstains when unsure
Low-confidence queries return ABSTAIN with a reason.
Supersession tracking
Declared newer facts win over stale similar notes.
Provenance on results
Each result carries a verdict, confidence and source.
Local Lite edition
One SQLite file, no database server.
PostgreSQL edition
pgvector, authentication and immutable generations.
Local embeddings
The default embedder runs on your machine.

Good for

  • Give a coding agent a project memory it can trust
  • Retire outdated decisions without deleting history
  • Search notes and markdown by meaning with refusal on no match
  • Serve shared team memory from Postgres

Pricing

Open source

AGPL-3.0; a commercial licence is available.

Prices change. Check giulioder.github.io before you buy.

RE-call MCP Memory Server FAQ

What licence is it?

AGPL-3.0, with a commercial licence available.

Do I need a database server?

Not for Lite, which uses a local SQLite file; the PostgreSQL edition needs pgvector.

Does it call an LLM?

The default path needs no memory-layer LLM call; answer providers are opt-in.

Which Python versions work?

Python 3.11 and above, with 3.12 to 3.14 also supported.

Is RE-call MCP Memory Server free?

Yes. RE-call MCP Memory Server is open source, so you can use and self-host it for free.

How do I add RE-call MCP Memory Server to Claude Code?

Run this command in your terminal: claude mcp add re-call --env RECALL_SERVING_DSN='<RECALL_SERVING_DSN>' -- uvx recall-rag

What are the best alternatives to RE-call MCP Memory Server?

The closest alternatives to RE-call MCP Memory Server are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and Midplane.

Alternatives to RE-call MCP Memory Server

  1. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  2. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  3. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  4. Local-first long-term memory layer for AI agents, served over MCP

    Not checked yet
    • 5 tools
    • Remote
    • Open source
    • 1.3K GitHub stars
    • Knowledge and memory
  5. Midplane

    MCP server

    Postgres gateway for AI agents with table grants, column masking and approval for writes

    Runs locally
    • Open source
    • 5 GitHub stars
    • Databases
  6. note1

    MCP server

    MCP server giving AI assistants search and summaries across your note1 meetings

    • 4 tools
    • Remote
    • Free trial
    • 1 GitHub star
    • Productivity