ShipSignal · AI impact audit

jest

14 May 2014 – 24 Jun 2026632.1 weeks7248 dev commitsexcluded 259 merges + 166 maintenance-bot
Generated 18 Jul 2026, 21:14 UTC · shipsignal v0.9.0
AI Adoption
None0%
Claude 1, Cody 1 · lower bound · → Emerging 0.2% recovered
Delivery Health
96/100 A
general eng norms
Readiness
not run
How to read this report
AI Adoption
The share of commits an AI tool co-authored — the one directly measured sign AI is actually being used here, not a survey.
Delivery Health
How sound the team's shipping habits are by general engineering norms — deliberately NOT credited to AI. High adoption means little if delivery health is poor.
Outcomes
How often changes get reverted or fixed, and how fast — outcome signals to complement the habit-based numbers above. Always context, never part of any score.
Release cadence
How often the repo tags releases, and how long a commit waits between landing and its release tag — deploy-frequency and lead-time proxies from data already in the clone. Always context, never part of any score.
Readiness
Whether the repo is set up so an AI agent (or a new human) can navigate it and trust what it reads — the conditions that decide whether AI adoption actually pays off.
Before/after AI Enablement
When a clean pre-AI baseline exists, how delivery metrics shifted after adoption — shown as context, never proof AI caused the change.
Trajectory
How AI adoption and delivery health moved over the repo's history — two parallel timelines, correlation only, never proof one caused the other.
AI adoption 0.0% (2/7248 commits — lower bound)
No sustained adoption window detected.
Breadth: 0% — 2 of 1766 active contributors · flat
Team-level only — ShipSignal does not score individuals.
Rate / week: ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▃▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ last 60w · 0–100%
The share of commits an AI tool co-authored — the one directly measured sign AI is actually being used here, not a survey.
↑ recovered Emerging 0.2% from PR data — +9 squash commit(s) re-attributed (Claude, Cody, Copilot, Cursor); measured 0% · 4782/5429 matched · coverage 88% — partial export, so the recovered figure is itself a lower bound

Delivery Health

How sound the team's shipping habits are by general engineering norms — deliberately NOT credited to AI. High adoption means little if delivery health is poor.

change_size_discipline
90%
test_discipline
100%
knowledge_distribution
100%

Context (not scored): 11.47 commits/wk · 1766 contributors.

Outcomes context, never scored

Revert pairs: 27 · median time-to-correction 0d (19 unmatched)

Change-failure proxy: 16% (1180 commits)

Release cadence context, never scored

Cadence: 1.35 tags/mo · median gap 5.6d (trailing 12 months, 252 tags)

Lead time: median 13.3d (7421 commits)

Before/after AI Enablement (bonus)

n/a — no adoption date found
A before/after needs a clean pre-AI baseline; the three numbers above stand on their own.

Trajectory over time — parallel timelines, NOT a causal link

0501002014-05-142025-08-25adoption %delivery health
Attribution caveat. Delivery pillars (flow, quality, risk) measure GENERAL delivery health — only AI-adoption and readiness are AI-specific. A delivery change may come from hiring, a finished migration, or a calmer quarter. The score asks whether the conditions under which AI pays off are improving — it does NOT prove AI caused any change.