Check product changes before you merge
Make agent evaluations part of your pull request review. The Gauge GitHub App runs realistic developer tasks against your candidate CLI, package, docs preview, or Skills and reports the outcome on the PR. See whether agents can use the changed product, which criteria failed, and the session evidence behind the result.Set up PR checks
Test your docs, Skills, CLI, or marketing site in real coding agent sessions
before you publish to production. Connect your repository and add eval cases
directly in code.
Measure agent behavior
Use Gauge to measure two parts of agent-led growth:- Agent Preference shows how often agents recommend or adopt your product for a realistic developer task.
- Agent Experience checks whether agents can complete a task with your product and why they pass or fail.
llms.txt, and docs, or improve agent experience with your CLI, Skills, and docs. To give an eval a private application as its starting workspace, see Run evals on private repositories.
Get started
Use the full Gauge Agents web product or the Gauge CLI to onboard, configure measurements and their run settings, launch runs, and inspect results. Both interfaces provide the same core Coding Agent optimization workflows. To onboard from your terminal, install the CLI and run the guided setup:How Gauge works
Follow a measurement from its prompt and run configuration through
execution, judging, and analysis.
CLI guide
Configure measurements, launch runs, and query results from your terminal or
CI.
API reference
Integrate with the bearer-token API that powers the Gauge CLI.