The AI-First Audit

See how AI-first your team really is.

Connect GitHub. We grade your team's shipping velocity, review hygiene, and PR cycle-time against the AI-first cohort — in two minutes.

2 minutes · Read-only · No code accessed · Revoke anytime
Read-only Metadata only Revocable anytime
Example Report
Acme Engineering · 47 engineers Last 90 days
B
78/100 AI-FIRST SCORE

Your team is shipping well, but the AI-first cohort has pulled ahead.

YOU
TOP 8%
The supporting metrics.
Median PR cycle time
2.4d +38%
Median time-to-first-review
3h 12m −14%
PRs per engineer per week
4.1 −22%
Stale PR ratio
31% +2.4x
Built for security review

Is this safe?

Engineering leaders should not have to take a marketing site at its word. Here is exactly what we access, what we don't, and how to revoke it in one click.

What we actually access.

We only read your repo metadata — PR titles, timestamps, reviewer counts, merge events, and the names of a short list of AI-convention files. We never read your code. We never write anything to your repos. You can revoke access in your GitHub settings any time.

  • Read-only access — we cannot modify your repos
  • Metadata only — PR titles, timestamps, reviewer counts
  • We never read your source code
  • Revocable in one click from GitHub settings
Technical details — for security review

What we read and what we don't

We detect the presence of a short list of AI-convention files in your repos (e.g. CLAUDE.md, AGENTS.md, .cursorrules). We only look at whether they exist — we don't read their contents.

What we never access:

  • Your source code
  • Branch contents
  • File diffs beyond filenames
  • Issue bodies and comments
  • Secrets, environment variables, deploy keys
  • Wiki pages, releases, packages

Data retention

Finished grade reports are held in an in-memory cache for 24 hours, keyed by job ID, then automatically discarded. Reports are never written to disk or to our database.

To request deletion, email [email protected].

Revocation

What you get

A report card you can actually use.

A letter grade

Honest, benchmarked against the AI-first cohort. No vanity metrics, no scorecards designed to make you feel good.

The supporting metrics

Cycle time, review latency, PR size, stale PR ratio — each shown with how the cohort performs.

A talking-points brief

Ready-to-paste paragraphs for your next leadership conversation. Frame the gap, not just the score.

How it works

Two minutes, start to score.

01

Connect GitHub

Read-only OAuth. We list your repos — you pick the ones to grade.

~30 seconds
02

We analyze

We pull your recent PR metadata and compute your score against the AI-first cohort.

~90 seconds
03

Get your grade

Report card delivered in your browser. Shareable URL for the team.

Instant

Ready to find out?

Two minutes. Read-only. No code accessed. Revoke anytime.