Developer docs

CI/CD Integration

Run automated QA checks as part of your deployment pipeline. One POST request returns a score, issue list, and pass/fail result — so you can block bad deploys before your users see them.

Authentication

Generate a secret key and set it as WEBHOOK_API_KEY in your Vercel environment variables. Pass the same value in every request.

# Option A — x-api-key header (recommended)
x-api-key: your_secret_key

# Option B — Authorization header
Authorization: Bearer your_secret_key

Endpoint

POSThttps://agentqa.viyalabs.com/api/webhook/scan

Request body

Field	Type	Required	Description
url	string	yes	The URL to scan (must be publicly accessible)
failThreshold	number	no	Minimum passing score 0–100. Default: `70`

Response

Returns 200 when score >= failThreshold, 422 otherwise. Your CI runner treats 422 as a build failure.

{
  "passed": true,
  "score": 87,
  "failThreshold": 75,
  "scanId": "3fa85f64-...",
  "url": "https://your-app.vercel.app",
  "reportUrl": "https://agentqa.viyalabs.com/report/3fa85f64-...",
  "summary": {
    "totalPages": 8,
    "totalIssues": 3,
    "critical": 0,
    "medium": 2,
    "low": 1
  },
  "criticalIssues": []
}

Status	Meaning
200	Scan passed — score ≥ failThreshold
401	Invalid or missing API key
422	Scan failed — score < failThreshold (also used for invalid URL)
429	Scanner busy — retry in a few minutes
500	Internal error

Examples

curl

curl -f -X POST https://agentqa.viyalabs.com/api/webhook/scan \
  -H "x-api-key: $AGENTQA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://your-app.vercel.app","failThreshold":75}'

GitHub Actions

# .github/workflows/qa.yml
name: QA Gate

on:
  deployment_status:

jobs:
  qa:
    if: github.event.deployment_status.state == 'success'
    runs-on: ubuntu-latest
    steps:
      - name: Run QA scan
        run: |
          curl -f -X POST https://agentqa.viyalabs.com/api/webhook/scan \
            -H "x-api-key: ${{ secrets.AGENTQA_API_KEY }}" \
            -H "Content-Type: application/json" \
            -d '{"url":"${{ github.event.deployment_status.target_url }}","failThreshold":75}'

Vercel + GitHub

# Use Vercel's deployment URL from the CLI output
PREVIEW_URL=$(vercel deploy --token $VERCEL_TOKEN)

curl -f -X POST https://agentqa.viyalabs.com/api/webhook/scan \
  -H "x-api-key: $AGENTQA_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{"url":"$PREVIEW_URL","failThreshold":80}"

Quick setup

1Generate a secret key — e.g. run: openssl rand -hex 32
2Add WEBHOOK_API_KEY=<your-key> to your Vercel environment variables
3Add AGENTQA_API_KEY as a secret in your GitHub repository (Settings → Secrets)
4Paste the GitHub Actions YAML above into .github/workflows/qa.yml
5Push — scans will run automatically on every successful deployment

What we detect

Every scan checks for the following issue types across all discovered pages. Issues are classified into three severity levels that feed directly into the QA score.

Type	Severity	Description
page_crash	critical	Page threw an unhandled exception during navigation — users see a blank or broken page
navigation_failure	critical	Playwright could not reach the page at all — server timeout, DNS failure, or hard crash
js_error	critical	Uncaught JavaScript exception detected in the browser console with full stack trace
page_not_found	medium	Page returned HTTP 404 — internal link pointing to a missing route
console_error	medium	Non-fatal console.error() call — often a failed resource, API error, or React warning
network_failure	medium	An XHR or fetch request returned a 4xx/5xx status or timed out
missing_image	medium	An <img> tag failed to load — broken src URL or missing file
missing_alt	medium	Images without alt text — WCAG 2.1 SC 1.1.1 accessibility violation
mobile_layout	medium	Content overflows the 375 px mobile viewport — users scroll horizontally
broken_form	medium	Form submission returned a network error or the form could not be submitted
slow_load	low	Page took longer than 5 seconds to reach interactive — performance regression risk
large_asset	low	A script, stylesheet, or image exceeded 500 KB uncompressed
console_warning	low	console.warn() calls — deprecation notices, missing keys, or React prop warnings
missing_meta	low	Missing meta description, Open Graph image, H1 heading, or mobile viewport tag

Score formula

The QA score starts at 100 and deducts points by severity. Deductions per severity are capped so a single category can't zero out your score alone.

Critical

−20 pts each

max −60

Medium

−8 pts each

max −30

Low

−2 pts each

max −10

Example: 1 critical + 2 medium + 3 low issues → 100 − 20 − 16 − 6 = 58 / 100. The failThreshold defaults to 70 in the CI/CD webhook — scores below that return HTTP 422.

Questions? info@viyalabs.com← Back to AgentQA