CI usage

Fail the job on red (exit 1). Exit 2 means no agent, not signed in, or unscored quality. Pass agent keys explicitly. Interactive skillgate login is for your laptop, not CI. No Skill Gate account required.

GitHub Actions

Point path at the skill pack folder (the directory that contains SKILL.md). In a workspace repo, that is often skills/<name>, not the repository root unless the pack lives there. Pass agent and decision-model keys as SKILLGATE_* variables. judge is the decision model id. The action does not switch to another decision model. Put --no-judge in args when the job should not call a decision model.

yaml

name: skillgate
on: [push, pull_request]
jobs:
  gate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: vigneshm1995personal/skillgate@v1
        with:
          path: skills/pdf
          judge: jev
        env:
          SKILLGATE_JEV_API_KEY: ${{ secrets.SKILLGATE_JEV_API_KEY }}

version pins the skillgate-ai release. Leave it unset for the latest release. model, base-url, and account-id are optional. harness is optional. args holds extra flags such as --expect-fail. The action does not print key values. Exit 0 passes, exit 1 fails, and exit 2 means the check cannot run. The job summary records that result.

Intentional fail suites

shell

skillgate example pdf-sabotaged --dest skills/pdf-sabotaged
skillgate check skills/pdf-sabotaged --expect-fail --key-env CURSOR_API_KEY --no-judge