Extelligence-ai/bagel

git clone https://github.com/Extelligence-ai/bagel.git && cd bagel
docker compose run --service-ports ros2-kilted

Usage

• •Start the Docker service matching your data format (ros2-kilted, ros2-jazzy, px4, ardupilot, betaflight, iot, etc.) and wait for Uvicorn on 0.0.0.0:8000.
• •Connect your LLM: claude mcp add --transport sse bagel http://0.0.0.0:8000/sse.
• •Ask in plain language, e.g. 'Summarize the metadata of the ROS2 bag ./data/sample/ros2/mcap'.
• •For fully offline use, pair with Ollama: uvx ollmcp --mcp-server-url http://localhost:8000/sse --model qwen3:8b.

Key features

• ✓Plain-language querying over ROS1/ROS2/MCAP, PX4, ArduPilot, Betaflight, ASAM MDF4/CAN, MQTT, PostgreSQL/TimescaleDB, and InfluxDB 3.
• ✓Deterministic DuckDB SQL with auditable queries displayed to the user — no black-box LLM math.
• ✓Natural-language pipelines: previewed before a byte is written, then run once, across a fleet, or standing at the edge.
• ✓Extensible via POML files ('tricks'); Claude Code plugin with four skills and runbooks for Gemini CLI, Codex, Cursor, and Copilot.
• ✓Integrations with Rerun, Lichtblick/Foxglove, PlotJuggler, Cloudini, Slack, and LeRobot (beta).

Best for

Engineers interrogating robotics/IoT/automotive log data conversationally instead of scripting per-format tools.

Caveats

• ⚠Requires Docker Desktop and an MCP-enabled LLM (Claude Code or another).
• ⚠Automotive MDF4/CAN and LeRobot integrations are marked beta.
• ⚠To give Bagel access to local files you must edit compose.yaml to uncomment volumes before starting.
• ⚠Not offline by default — the LLM connects remotely unless you bring a local model.