Chronos MCP

Gives Claude a long-term memory and the ability to map out your projects.

Claude forgets everything when you start a new chat. You end up pasting the same context, rules, and project states over and over. Chronos fixes this by giving Claude a permanent, local memory it can search — and a structured graph it can use to track complex projects. Everything stays on your machine. No accounts, no APIs, no cloud.

Quick Start

1. Install

git clone https://github.com/you/ChronosMCP
cd ChronosMCP
pip install -e .

2. Add to Claude Desktop

Open claude_desktop_config.json and add:

{
  "mcpServers": {
    "chronos": {
      "command": "python",
      // Path to wherever you cloned ChronosMCP
      "args": ["/absolute/path/to/ChronosMCP/chronos_mcp.py"],
      "env": {
        // Where Chronos saves your database. Defaults to chronos.db next to the script.
        "CHRONOS_DB_PATH": "/absolute/path/to/chronos.db"
      }
    }
  }
}

Claude Code (CLI):

claude mcp add chronos -- python /absolute/path/to/ChronosMCP/chronos_mcp.py

Restart Claude Desktop and you're done.

What You Can Ask Claude

Remember that we always squash commits before merging to main.

What do you know about our Git workflow?

What did you know about the auth system as of March 1st?

Forget the note about the old staging URL.

What tasks are blocking the payments feature right now?

What should I work on next in the api-rewrite project?

Tools

Chronos gives Claude two distinct pipelines: a memory layer for free-text notes, and a graph layer for structured project tracking.

Memory

Graph

Configuration

How It Works

When you ask Claude to remember something, Chronos writes it to a local SQLite file and indexes it using TF-IDF — no dependencies, no model downloads required. recall ranks results by relevance and applies a gentle recency boost so newer memories surface first without burying highly relevant older ones.

For project tracking, Chronos organizes nodes in a hyperbolic space rather than a flat list. This means closely related tasks cluster naturally together even as your graph grows — the geometry handles the relationships so you don't have to label everything manually.

query_at time-travels by reconstructing a snapshot of your memory database at any past timestamp, including resolving content that's since been edited. The database is a single local file. Back it up however you back up files.

License

MIT