EVIDENCE DASHBOARD

Scheduled evidence, auditable.

CI conclusions, median runtimes, releases, and repository activity for five reference frameworks — pulled from public GitHub data on a daily schedule, with explicit provenance and stale/unavailable states, readable without JavaScript.

Scheduled evidence snapshot 02/08/2026, 06:47:31 UTC
VERITY PLAYWRIGHT K6 ARIA SELENIUM

START HERE

Recommended review order

Five repositories, one layered quality model. Each has a documented, time-boxed review path — this is the order that tells the story fastest.

  1. Verity Policy Coverage Eval12 minute review
  2. Playwright TypeScript12 minute review
  3. k6 Performance12 minute review
  4. ARIA API10 minute review
  5. Selenium TestNG Java10 minute review

REPOSITORY MATRIX

Evidence at a glance

Latest CI conclusion, median workflow runtime over the last ten completed runs, current release, and default-branch activity — per framework.

Repository Status Evidence state Latest CI Median runtime Release Activity Evidence
Verity Policy Coverage Eval Configured evidence-unavailable success 2m 07s (n=10) v0.1.0 358 commits · 29 PRs Verification Limitations Report Docs Screenshot
Playwright TypeScript Configured evidence-unavailable failure 3m 13s (n=10) v1.0.0 48 commits · 47 PRs Verification Limitations Report Docs Screenshot
ARIA API Configured evidence-unavailable failure 2m 54s (n=10) v1.0.0 58 commits · 65 PRs Verification Limitations Report Docs Screenshot
Selenium TestNG Java Configured evidence-unavailable success 4m 54s (n=10) v1.0.0 79 commits · 33 PRs Verification Limitations Report Docs Screenshot
k6 Performance Configured evidence-unavailable success 8m 55s (n=10) v0.4.0 139 commits · 26 PRs Verification Limitations Report Docs Screenshot

PLATFORM CASE STUDY

Scaling shared automation with reliability controls

An anonymized client migration, ≈1 month: replace raw test count with a layered pyramid of critical journeys plus API coverage.

SUITE STABILITY

~40% → ~90%

before → after

FULL-SUITE RUNTIME

~2 h → ~15 min

8× faster feedback

TEST MIX

1,000+ → 40 + API

UI-heavy checks → critical journeys + API

AUTHORING SPEED

~3× faster

fixtures · builders · conventions

Problem
Suite scale outgrew manual ownership and serial feedback.
Decision
Stop optimizing for raw test count; make each layer own a distinct failure signal.
Controls
Layer allocation, parallelism, ownership, release policy, and triage standards.
Evidence rule
Professional outcomes are labeled separately from public repository metrics.

Read the full case study