CAI Get a survey →

The open standard

An open, reproducible 0–100 standard for the health of a codebase.

Same code in, same score out — a measurement anyone can verify. The standard is composed of 127 dimensions across 10 lenses, each pinned to a versioned rubric. The method is open; the independent, signed survey is a service.

Read the spec Browse the 127 dimensions Score an evidence bundle →

Open · Verifiable · Referenced

Open

The full catalog — 127 dimensions, 10 lenses — and the fold are public and versioned. A reference scorer ships under Apache-2.0.

Verifiable

The score is deterministic. Same evidence + same rubric version ⇒ the same number — and you can run it yourself.

Referenced

This site is the authority: the rubric catalogs are served by API, so a survey, a contract, or a badge points back to a checkable record.

The bands

Exemplary 90–100 Strong 70–89 Adequate 50–69 Weak 25–49 Critical 0–24

The firewall — what is open, and what is judgment

The score, the findings, and the algorithm are the open standard: deterministic and reproducible. The surveyor's deductions and advice sit outside the firewall and never move the number. That boundary is also the free/paid line.

The open standard — deterministic same evidence + same rubric version → the same number, reproducible by anyone CAI 0–100 — the headline (never moves) ↑ worst-first ordered-weighted fold of the lenses the lenses — each rolls up dimensions scored 0–10 by deterministic tools advisory deductions & advice (never scores)
Anyone can recompute the deterministic score from the open evidence; the advisory layer explains but never scores. Openness makes independence a fact you can check, not a claim.

The standard is free. The judgment is a service.

Use CAI to measure, compare, and verify any codebase yourself. When you need an independent, signed survey — the CAI plus the deductions and what to do about them — get one from the surveyor.

Get an independent survey →