Krabby

[](https://raw.githubusercontent.com/rytsh/krabby/main/LICENSE) [](https://sonarcloud.io/summary/overall?id=rytsh_krabby) [](https://github.com/rytsh/krabby/actions) [](https://github.com/rytsh/krabby/pkgs/container/krabby) [](https://rytsh.github.io/krabby/)

Krabby indexes your repositories and lets AI agents search code, read documentation, and understand relationships through MCP.

Features

• Repository indexing with automatic background refreshes
• Fast code search and file access
• Architecture and relationship analysis with knowledge graphs
• Optional generated documentation and semantic search
• Web, Confluence, and Jira source indexing
• Web UI, REST API, and MCP support

Run with Docker

With Docker Compose:

docker compose --project-name krabby up -d --pull always

To stop and remove Krabby while keeping its data:

docker compose --project-name krabby down

# Also permanently delete all Krabby data:
# docker compose --project-name krabby down --volumes

Or with Docker directly:

mkdir -p krabby-data

docker run -d \
  --name krabby \
  -p 8080:8080 \
  -v "$(pwd)/krabby-data:/data" \
  ghcr.io/rytsh/krabby:latest

Open http://localhost:8080, then add repositories from the UI. The krabby-data volume (or directory when using docker run) keeps repositories, indexes, and settings between container restarts.

> For base path use KRABBYSERVERBASE_PATH environment variable, e.g. -e KRABBY_SERVER_BASE_PATH=/krabby to run behind a reverse proxy.

Add MCP

Go to About section in the UI and follow the instructions to add Krabby to your MCP client configuration.

Click for MCP configuration

Krabby exposes a streamable HTTP MCP endpoint at http://localhost:8080/mcp.

Paste this into your coding agent:

> Add Krabby to my MCP client configuration as a remote streamable HTTP server. > Name it "krabby", use http://localhost:8080/mcp as the URL, and send the > X-Krabby-Tool-Profile: full header on every request. Preserve my existing MCP > servers and tell me if I need to restart the client.

The full profile lets your agent configure Krabby and manage credentials in addition to adding repositories, searching code, reading files, and querying repository relationships.

For client-specific configuration examples, private repository credentials, REST endpoints, MCP tools, memory tuning, and development instructions, see [DETAILS.md](DETAILS.md).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: rytsh
• Source: rytsh/krabby
• License: MIT
• Homepage: https://rytsh.io/krabby/

Install and usage instructions live in the source repository linked above.