Run an automated consent-state check against a public website from GitHub Actions. ScanBell opens the page in fresh browser contexts, records the default state, attempts a reject path, conditionally checks acceptance, and returns confidence-labelled technical findings.
name: Consent QA
on:
workflow_dispatch:
deployment_status:
jobs:
scan:
if: github.event_name == 'workflow_dispatch' || github.event.deployment_status.state == 'success'
runs-on: ubuntu-latest
steps:
- uses: tmrp/scanbell-action@v1
with:
url: https://www.example.com
fail_on: neverfail_on defaults to never. Set it to high only when you want high or critical technical findings to fail the job. Findings are review signals rather than legal conclusions, so a non-blocking first run is recommended.
scan_id— the scan identifier.score— the technical QA score from 0 to 100.high_count— combined high and critical finding count.finding_count— total finding count.
The action also writes a compact report to the GitHub job summary.
The url is sent to the public ScanBell scanning service. The service scans only publicly reachable HTTP(S) pages, uses bounded fresh-browser sessions, and applies abuse and concurrency limits. It records aggregate product events and technical scan metrics; it does not store IP addresses, user agents, referrers, cookie values, or browser-storage values.
Do not use this action for private, authenticated, or sensitive URLs. Availability is best-effort while the free experiment is being validated.
The free check is designed for fast release QA. For three representative URLs, timestamped default/reject/accept evidence, remediation notes, a client-shareable report, and one verification rerun, review the full simulated deliverable, run ScanBell, or open a €49 evidence-pack request.
ScanBell reports automated technical observations. It is not legal advice and does not certify compliance with GDPR, ePrivacy rules, or any other law.
MIT