Answer
Is there a benchmark showing how often a coding agent builds a working CRM with this?
No. The build benchmark protocol is designed and published, but it has not been run, so no Successful Agent Build Rate exists and any number quoted for this project is not ours (L-03). What exists instead is a test suite: 701 tests run on every push, covering happy paths and the policy boundaries that matter — hostile input, transaction rollback, idempotency, concurrency and immutability among them (C-20). That count measures effort, not correctness, and real-browser tests are run manually and are not in CI.
What the ledger says, word for word
Each entry below is copied from site/claims.json, where it is bound to the evidence that holds it — a test file, a document, or a named repository fact — and to the limitation that travels with it. The answer above summarises these; these are the claim.
C-20
701 tests, run on every push, covering happy paths and the policy boundaries that matter — hostile input, transaction rollback, idempotency, concurrency and immutability among them.
LimitA test count measures effort, not correctness — read the adversarial-review categories in docs/QUALITY_GATES.md to see what is actually attacked. Real-browser tests are run manually and are not in CI.
docs/QUALITY_GATES.mdnpm run verify at 9958ed9: 701 passing, 0 failing.github/workflows/ci.yml runs verify and smoke on push and pull_requestL-03 · The build benchmark has not been run.
The protocol is designed and published; no Successful Agent Build Rate exists yet. Any number you see quoted for this project is not ours.
docs/strategy/CRM_BUILD_BENCHMARK.mddocs/PROJECT_STATUS.mdWhere this stops
Nothing on this page implies the framework is deployable. There is no authentication, tenancy or RBAC, so it is local-development-only whatever any single answer says. Every claim and every limitation is on one page, and the questions this project refuses to answer are published beside them.