Catch agent regressions before you merge.

Every change to your SDK, CLI, skills, or docs can break the path agents take. Gauge runs coding agent evaluations on each pull request and posts the result before you merge.

Results on the pull request

The Gauge GitHub App posts pass or fail results and session links on the PR.

Uses your build artifacts

Test candidate packages, executables, skills, files, or docs previews from the builds and previews you already have.

Works with your CI

Use the GitHub App, or trigger hosted runs from your own CI.

FAQs

Agent Experience checks evaluate implementation success against your criteria. Agent Preference checks report product selection as advisory feedback.

Yes. Gauge routes requests for your production docs origin to the PR’s preview inside the sandbox. The preview must be publicly accessible and associated with the candidate commit.

Gauge supports Claude Code, Codex, OpenCode, Pi, and Antigravity. The hosted Free plan includes open models on OpenCode and a first onboarding run on Claude Code and Codex at no cost. Upgrade to Pro or Enterprise for ongoing Claude Code and Codex runs and access to all supported agents.