Foxglove3.3.0

Foxglove lets robotics teams collect, inspect, and debug fleet recordings, then locate important moments for investigation or model development. It brings scattered robot data into a queryable workspace for physical-AI work, helping engineers understand behavior across individual runs and entire fleets.

Teams can record logs in the open MCAP format with Foxglove’s C++, Python, or Rust SDKs, while Foxlet on a robot can automatically upload ROS 1 bag and MCAP recordings. Imported data is indexed by device, time, and topic, and recordings in MCAP, ROS Bag, ULog, and other formats can be viewed locally, from an organization’s cloud data platform, or through remote HTTP servers. The Foxglove app also connects to robots for live visualization or teleoperation and can compare multiple runs to reveal behavioral differences.

Engineers can search all recordings for noteworthy moments, export the results, and share recordings or events with teammates handling an investigation. Events and metadata can be added to logs for easier discovery, while built-in agents or agents connected through MCP can help debug data and configure visualization layouts.

Foxglove suits robotics developers working on autonomy, fleet operations, and training datasets, especially when diagnosing a physical system or curating examples for a new model. The current release is 3.3.0 for macOS 12 Monterey or later on Apple Silicon (arm64), and it is freeware that is free to start.