GitHub Action
Gate your repository’s Agent Skills (SKILL.md) in CI: fail on security findings
(prompt injection, hidden Unicode, dangerous commands, credential leaks, exfiltration,
dangerous scripts) and on any drift from the approved skillwarden.lock baseline.
name: Skill Gateon: push: branches: [main] pull_request:
jobs: skillwarden: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - uses: wookat/skillwarden/packages/action@main with: fail-on: high # low | medium | high | criticalInputs
Section titled “Inputs”| Input | Default | Description |
|---|---|---|
command |
ci |
ci (drift gate + scan), scan, or diff |
paths |
auto-discover | Space-separated skill paths |
fail-on |
high |
Severity threshold that fails the gate |
lockfile |
skillwarden.lock |
Lockfile path |
version |
latest |
skillwarden npm version to run |
Requires Node.js 22+ on the runner (present on ubuntu-latest).
Typical workflow
Section titled “Typical workflow”- Review the skills in your repo, then commit a baseline:
skillwarden lock && git add skillwarden.lock. - Add the workflow above. From then on, any upstream skill change (drift) or any
finding at/above
fail-onturns the build red. - When a change is legitimate, re-approve with
skillwarden lockand commit the updated lockfile.