ai-dev-tools-debug

You can debug AI evaluation failures by accessing repositories, eval data, artifacts, and logs through the CLI and app workflows. This skill provides guided discovery and read-only inspection to help you reproduce issues and identify root causes across multiple environments.

The best way to reproduce a reported bug is to list repositories, locate the target repo paths, and inspect artifacts and plan logs. This structured workflow verifies fixes and reproduces failures safely without altering user data.

Yes, you can inspect plan logs and artifacts without modifying user data. This skill provides safe, read-only guidance with CLI commands and checks to diagnose problems across eval systems, plan runners, and the macOS app.

Troubleshoot issues across multiple repositories and environments by using coordinated discovery and read-only inspection workflows. The skill exposes access to eval cases, artifacts, and log data to help you debug across the eval system, plan runner, and macOS app.

Yes, this debugging workflow supports the macOS app and plan runner. It provides guided discovery and debugging workflows across eval systems, the plan runner, and the macOS app to help you reproduce user-reported bugs and verify fixes.

The limitation of read-only CLI debugging for AI evaluations is that it provides safe, read-only guidance to diagnose problems without altering user data. It coordinates discovery and inspection but does not directly apply fixes or modify repository state.