Skip to content
MCP server

LLMSlim memory

Local long-term memory for agents that returns the relevant context within a token budget

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
CLI, self-hosted, API
Runs
Locally
Version
0.12.0
Listing updated
3 days ago

About LLMSlim memory

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

LLMSlim memory stores what users and agents say word for word in a local SQLite file, then returns the parts that answer a question within a token budget you set. No language model is called in the loop, so there are no API keys or network calls for the default extractive strategy.

The same engine runs as this MCP server, a Python library or a small HTTP server, and the README gives setup commands for Claude Code, Codex, GitHub Copilot CLI, Gemini CLI and a Hermes Agent plugin. Environment variables choose the database file, whose memory the server holds and an optional agent id so each agent keeps its own memory. Postgres and MongoDB backends are available through extras.

The core is MIT-licensed and free. The site describes a planned hosted edition called Orbit, expected to open with version 1.0, and a self-hosted Vault in private beta.

Install

Run in a terminal

claude mcp add llmslim -- uvx llmslim

Add to claude_desktop_config.json

{
  "mcpServers": {
    "llmslim": {
      "command": "uvx",
      "args": [
        "llmslim"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "llmslim": {
      "command": "uvx",
      "args": [
        "llmslim"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "llmslim": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "llmslim"
      ]
    }
  }
}

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

{
  "mcpServers": {
    "llmslim": {
      "command": "uvx",
      "args": [
        "llmslim"
      ]
    }
  }
}

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

{
  "mcpServers": {
    "llmslim": {
      "command": "uvx",
      "args": [
        "llmslim"
      ]
    }
  }
}

Configuration

  • LLMSLIM_MEMORY_DBSQLite file for the memory (default ~/.llmslim/memory.db)
  • LLMSLIM_USER_IDWhose memory this server holds (default "default")
  • LLMSLIM_AGENT_IDOptional agent id, so each agent keeps its own memory

What it does

No LLM in the loop
Selection and compression run locally without a model call.
Token-budgeted recall
Search returns the passages that answer a question within your budget.
Per-agent memory
Set a user id and agent id so memories stay separate.
Local SQLite storage
Memory lives in a file, by default ~/.llmslim/memory.db.
Library and HTTP server
Use it from Python, TypeScript or a small server.
Framework adapters
Adapters exist for LangChain, LlamaIndex, Haystack, gptme and Pydantic AI.

Good for

  • Give a coding agent memory that persists between sessions
  • Fit long chat histories into a fixed token budget
  • Keep separate memories for several agents
  • Compress retrieved documents before a model call
  • Run agent memory offline

Pricing

Open source

Core is free (MIT). Hosted Orbit plans planned, from $19/month.

Prices change. Check llmslim.app before you buy.

LLMSlim memory FAQ

Does it need an API key?

No. The default extractive strategy runs in your process without one.

Where is memory stored?

In SQLite at ~/.llmslim/memory.db unless LLMSLIM_MEMORY_DB points elsewhere.

What licence is the core under?

MIT.

Is there a hosted version?

The site describes Orbit as opening with LLMSlim 1.0, planned for 8 December 2026.

Is LLMSlim memory free?

Yes. LLMSlim memory is open source, so you can use and self-host it for free.

How do I add LLMSlim memory to Claude Code?

Run this command in your terminal: claude mcp add llmslim -- uvx llmslim

What are the best alternatives to LLMSlim memory?

The closest alternatives to LLMSlim memory are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

Alternatives to LLMSlim memory

  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. note1

    MCP server

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

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