Gate deploys on SEO and GEO health with crawlie's non-zero exit codes — a ready-to-paste GitHub Actions workflow, thresholds, and agent-driven checks.

crawlie is built to run unattended. --fail-on makes it a quality gate: it exits non-zero when findings at or above a severity exist, which fails the job.

GitHub Actions

name: SEO gate
on: [deployment_status]

jobs:
  crawlie:
    runs-on: ubuntu-latest
    timeout-minutes: 15
    steps:
      - uses: actions/setup-node@v4
        with: { node-version: 20 }
      - name: Audit the deploy
        run: npx -y crawlie crawl "$SITE_URL" --fail-on error --quiet
        env:
          SITE_URL: ${{ github.event.deployment_status.target_url }}

Pin a version for reproducible runs: crawlie@0.2.0.

Thresholds

`--fail-on`	Fails when…
`none` (default)	never — informational only
`error`	any error-severity finding exists
`warning`	any error or warning finding exists

Pre-launch checks for agents

If you drive CI through an agent, the pre-launch-seo-check skill wraps this into a go / no-go verdict with the exact blockers listed. The agent runs the same --fail-on semantics and reports what to fix before you ship.

GitLab CI

The same idea works in any pipeline — the non-zero exit code fails the job:

seo_gate:
  image: node:20
  script:
    - npx -y crawlie crawl "$SITE_URL" --fail-on error --quiet

How do I read the result in CI?

The exit code is the gate: 0 means no findings at or above your threshold, non-zero means the build should stop. For detail, add --format json -o report.json and archive it as a build artifact — the JSON includes the Health and GEO scores, every finding with its severity and rule id, and per-page data, so you can diff runs or surface a summary in the job log.

Should I gate on errors or warnings?

Start with --fail-on error. Errors are unambiguous breakage — broken links, 5xx pages, missing titles on indexable pages — and rarely produce false positives, so they make a safe blocking gate. Treat warnings and notices as informational at first; once your site is consistently clean, tighten to --fail-on warning to hold the higher bar.

Tips

Audit a few critical pages instead of a full crawl for a fast gate: crawlie audit "$HOME_URL" "$PRICING_URL" --quiet.
Save baselines with --save so you can compare runs over time.
Use --format json -o report.json to archive the full result as a build artifact.
Pin the version (crawlie@0.2.0) so a new release can’t change your gate’s behavior mid-pipeline.

CI & automation