Lighthouse Bridge

lighthouse-bridge - Lighthouse CI Data Bridge

Hidden helper agent that bridges Lighthouse CI accessibility audit data into the agent ecosystem. Parses Lighthouse reports, normalizes findings, tracks score regressions, and deduplicates against local scans.

When to Use It

You do not invoke this agent directly. It is called automatically by:

• web-accessibility-wizard -- during Phase 0 (auto-detection) and Phase 9 (correlation)
• insiders-a11y-tracker -- when discovering Lighthouse-related issues and score regressions
• daily-briefing -- when collecting CI scanner accessibility updates

What It Does

1. Detects Lighthouse CI configuration -- Reads .github/workflows/ for treosh/lighthouse-ci-action and checks for lighthouserc.* config files
2. Parses Lighthouse reports -- Extracts accessibility score (0-100) and individual audit failures from Lighthouse JSON reports
3. Normalizes findings -- Maps Lighthouse weight-based severity to the standard model (Critical/Serious/Moderate/Minor)
4. Tracks score regressions -- Compares current vs previous accessibility scores, classifies delta as critical/serious/moderate regression or improvement
5. Deduplicates -- Correlates Lighthouse audit IDs with local axe-core rule IDs to avoid double-counting
6. Generates structured output -- Returns JSON with score, delta, findings, and regression status

Output Contract

The lighthouse-bridge returns structured JSON:

{
  "lighthouseDetected": true,
  "overallScore": 87,
  "previousScore": 95,
  "scoreDelta": -8,
  "totalFindings": 8,
  "bySeverity": {
    "critical": 1,
    "serious": 3,
    "moderate": 3,
    "minor": 1
  },
  "regressionStatus": "regressed-serious",
  "findings": [
    {
      "source": "lighthouse-ci",
      "ruleId": "color-contrast",
      "wcagCriterion": "1.4.3",
      "severity": "serious",
      "confidence": "medium",
      "url": "https://example.com",
      "element": ".header-text",
      "description": "Elements must meet minimum color contrast ratio thresholds",
      "lighthouseWeight": 7
    }
  ],
  "scoreHistory": [
    { "run": "2025-01-15", "score": 95 },
    { "run": "2025-01-22", "score": 87 }
  ]
}

Behavioral Rules

1. Read-only -- Never creates, edits, or closes issues. Only reads reports and returns data.
2. Structured output -- Always returns JSON matching the output contract.
3. Fail gracefully -- If no Lighthouse CI is configured or no reports are available, returns lighthouseDetected: false with empty findings.
4. Progress announcements -- Announces each phase of work as it proceeds.
5. No user interaction -- Never prompts the user. Works silently as a subagent.
6. Score context -- Always includes score context (previous score, delta, regression status) when available.

Platform Availability

Platform	File	Tools
GitHub Copilot	.github/agents/lighthouse-bridge.agent.md	github/*, fetch, readFile, textSearch
Claude Code	.claude/agents/lighthouse-bridge.md	Read, Grep, Glob, WebFetch, GitHub

Related

• Lighthouse CI Scanner Integration -- full setup and configuration guide
• lighthouse-scanner Skill -- knowledge domain reference
• web-accessibility-wizard -- the primary consumer of lighthouse-bridge data