Pentest Every Release — Without Slowing Down Deployments
Automated penetration testing (DAST) built into your CI/CD pipeline. Every release is attacked like a real attacker would, and the build fails only on confirmed vulnerabilities — each proven with the exact request and response.
CI/CD API on Professional and Business · GitHub Actions, GitLab CI, Jenkins, any CI
Quick answer
To pentest every release without slowing deployments, run a focused scan of your staging or preview URL in the pipeline (usually minutes) and fail the build only on confirmed findings above your chosen severity. Run full-domain scans nightly and get the result by webhook, so long scans never block a release.
How it fits into your pipeline
Verify your domain
Settings → Domains → add a DNS TXT record. Pipeline scans run unattended, so only verified domains can be scanned.
Create an API key
Settings → API Keys (Professional or Business). Store it as a CI secret named NAUTILLO_API_KEY.
Add one step to your pipeline
Use the GitHub Action, or two HTTP calls from any CI system: trigger the scan, then read the result.
Set the threshold
Fail the build on critical, high, medium or low confirmed findings — or report only, and get the full report link in the job output.
Examples
GitHub Actions — on every push to staging
- name: Security scan
uses: North-Human-AI/nautillo-pro-scan-action@v1
with:
api-key: ${{ secrets.NAUTILLO_API_KEY }}
url: https://staging.your-app.com
fail-on: highGitLab CI (works the same in any CI with curl + jq)
security_scan:
stage: test
image: alpine:3.20
before_script:
- apk add --no-cache curl jq
script:
- |
SCAN_ID=$(curl -sf -X POST https://api.nautillo.pro/functions/v1/v1-scan-trigger \
-H "Content-Type: application/json" -H "X-API-Key: $NAUTILLO_API_KEY" \
-d '{"target_url":"https://staging.your-app.com","fail_on":"high"}' | jq -r '.scan_id')
for i in $(seq 1 45); do
R=$(curl -sf -H "X-API-Key: $NAUTILLO_API_KEY" \
"https://api.nautillo.pro/functions/v1/v1-scan-status?scan_id=$SCAN_ID")
S=$(echo "$R" | jq -r '.status'); echo "[$i] $S"
[ "$S" = "completed" ] || [ "$S" = "failed" ] || [ "$S" = "timeout" ] && break
sleep 30
done
exit $(echo "$R" | jq -r '.exit_code // 2')Nightly full-domain scan with webhook
on:
schedule:
- cron: '0 2 * * *' # nightly full-domain scan
jobs:
full-domain-scan:
runs-on: ubuntu-latest
steps:
- uses: North-Human-AI/nautillo-pro-scan-action@v1
with:
api-key: ${{ secrets.NAUTILLO_API_KEY }}
url: https://your-app.com
scan-type: full_domain
timeout: '21600'
webhook-url: ${{ secrets.SCAN_WEBHOOK_URL }}Exit codes: 0 no confirmed findings at the threshold · 1 confirmed findings · 2 scan error or timeout. Full reference in the API docs.
Why test in the development cycle
Annual pentests were designed for software that changes once a year. Modern teams ship daily — and the gap between tests is where vulnerabilities live.
Tested on every release
Trigger a scan from your pipeline on every merge or deploy to staging. Security testing becomes part of the release, not a separate calendar event.
Fast enough for the pipeline
Focused single-URL scans usually finish in minutes. Long full-domain scans run nightly and report back by webhook, so they never block a deploy.
Fail the build on proof, not noise
Choose a severity threshold. The build fails only on confirmed findings — each one proven with the exact request and response — so developers aren't buried in false positives.
Signed webhooks
Receive a signed POST (HMAC-SHA256) when a scan completes — route it to Slack, a ticketing system or your own service.
Catch regressions
Every scan is saved. Compare runs to see which vulnerabilities are new, fixed or back again after a refactor or dependency update.
Authenticated scans from CI secrets
Test logged-in areas with a bearer token, basic auth or form login passed from your CI secrets. Credentials are encrypted and deleted after the scan.
Continuous testing vs periodic penetration testing
Periodic pentests still have their place — compliance mandates and deep manual review of business logic. Continuous testing covers everything in between, on every release.
| Dimension | Periodic pentest | Continuous testing in CI/CD |
|---|---|---|
| Test frequency | Once or twice a year | Every release, plus nightly |
| Gap between tests | 3–12 months | Hours |
| Time to finding | Days to weeks | Minutes for a focused scan |
| Cost per test | €4,000 – €25,000+ | Included in subscription |
| Developer feedback loop | Too slow to act on | Fixed while the code is fresh |
| Regressions | Not visible between tests | Compare saved runs |
| Blocks bad releases | No | Yes — fail the build on confirmed findings |
| Depth on complex business logic | Yes — human judgement | Partial — combine with a periodic pentest |
Frequently asked questions
Run a focused single-URL scan of your staging or preview environment in the pipeline — it usually finishes in minutes — and fail the build only on confirmed findings at the severity you choose. Run broader full-domain scans on a nightly schedule and receive the result by webhook, so long scans never block a deploy.
Put security testing in your next release
Start free to see the reports. Professional (€69/month) adds the full active attack suite and the CI/CD API with 5,000 triggers per day; Business adds unlimited triggers and deeper attack chains.