nominee

Proof

Case studies

None published. A vague customer story is worse than none — we will not ship a named (or anonymised) study without the partner's numbers and written sign-off.

Three slots are reserved. Each will include what they did before, what nominee caught (preferably from observe mode), how much adoption actually cost, and one thing that was harder than expected.

Status board and template: docs/case-studies.

Runnable evidence (not case studies)

These numbers come from this repository. They are not a named team's production metrics.

  • prompt-injection-blocked — a scripted agent obeys the injected forward; email.forward is denied before the mailer runs. The example does not call an LLM.
  • npx nominee-cli — $25 runs, $200 waits, $2,000 never reaches refund.issue.
  • token-refresh-correctness — naive rotating refresh fails 7/8 under concurrency; nominee 8/8.