Fix agent friction. Prove the fix.

Gauge groups what slowed agents down into issues ranked by impact. Edit the docs, skills, or CLI rules behind an issue, rerun the same sessions, and see what the change fixed before you ship it.

Issues ranked by impact

Findings from every session are grouped into issues, so you fix the problems that cost you the most runs first.

Same sessions, new inputs

Optimizations rerun the same task, harness, and model with copies of the pages, skills, or CLI rules you changed.

Resolved, returned, or introduced

Each original finding comes back resolved or returned, and new problems show as introduced.

FAQs

An issue groups the findings from many coding sessions that share a cause, such as a misleading install step or a page agents fetch again and again. Issues are ranked by impact, and you can mark them fixed or muted.

Your team or your coding agent. Gauge suggests changes based on the sessions, reruns the sessions with each change, and analyzes the result.

Publish the version you choose through your normal workflow, then rerun past sessions of the issue to verify the fix.