For capability publishers

Turn support anecdotes into scoped compatibility evidence

Capability Lab is launching the user-side diagnostic instrument first. Publisher monitoring becomes available only after enough reviewed outcomes exist to support environment-specific claims.

Test the user diagnostic

What the evidence model records

  • Capability and provider version
  • Codex surface and host version
  • OS/runtime/workspace context when relevant
  • First failing checkpoint
  • Safe verification and outcome
  • Source and review state

What it will not claim

  • No universal “works/doesn't work” verdict
  • No publisher score from one report
  • No public incident from an unreviewed anecdote
  • No execution of the publisher's package
  • No exposure of private user reports

Planned task chain

  1. Private diagnosis and outcome
  2. Reviewed, reproducible observation
  3. Environment-scoped incident record
  4. Compatibility matrix
  5. Release/regression monitoring

There is no publisher dashboard, paid plan, monitoring promise, or public compatibility score at launch. This page documents the evidence contract rather than advertising an unfinished product.