Lighthouse
Measured on the homepage. Scores below the threshold auto-open a GitHub issue tagged with the failing category. A score of – means the audit hasn't run yet.
–
Performance
Min: 70
–
Accessibility
Min: 80
–
Best Practices
Min: 80
–
SEO
Min: 90
Quality & Security
HTML validation runs html-proofer against the built site. Security headers measured by Mozilla Observatory. SSL/TLS measured by Qualys SSL Labs.
HTML & Links
–
html-proofer
Security Headers
–
Mozilla Observatory
SSL / TLS
–
SSL Labs
Open-source impact
Live counts pulled from the GitHub API across maintained repos — Provenance, virtualjaguar-libretro, Hero, RxRealm, XMLCoder, 240p Test Suite, and others. Updated on every status audit.
–
Stars across maintained repos
–
PRs merged (lifetime)
–
Provenance stars
–
Provenance downloads
–
Total contributors
–
Discord members
AI pipeline
Multi-agent CI ecosystem across the Provenance org. Three autonomous agents — Claude, Cursor, Kimi — share a dispatch vocabulary and route by branch / label. See the case study for architecture details.
3
Autonomous agents in production
–
AI-driven workflows
5
LLMs orchestrated
11
Languages auto-localized
Sonnet vs Opus auto-selection by static diff analysis
Cursor agent with shared dispatch vocabulary
Kimi Code as Anthropic-API-compatible fallback
Bot-actor gate bypass via scheduled cron
Diff-aware Claude Haiku localization (11 languages)
Cron-driven stuck-PR detection & re-dispatch
Claude self-heals spell-check & link-check failures
90+ / passing
at threshold
below threshold
not yet measured