Submission standards
Only submit what can survive scrutiny.
The Skills Library rejects decorative prompt bundles, misleading claims, undocumented side effects, or anything that creates the impression of certainty without evidence.
Clear purpose
What the skill does, who it is for, and what problems it solves.
Tested behaviour
Evidence that the skill produces correct output on representative inputs. Include test cases.
Known limits
Honest documentation of edge cases, failure modes, and situations where the skill should not be used.
Maintenance owner
A named contributor or operator responsible for updates, bug fixes, and quality maintenance.
Manifest file
A manifest.yaml with required metadata fields: name, version, author, licence, dependencies, and compatibility.
Production or provisional status
An explicit declaration of whether the skill is production-grade or still provisional.
Review process
Every submission is reviewed by at least one certified contributor. Reviews assess correctness, documentation quality, test coverage, and potential for misuse. Certification is revocable if quality degrades over time.
The submission flow is intentionally simple: documented asset, explicit review status, explicit maintenance owner, and a clear note on production readiness.
The Skills Library is not yet accepting submissions. When it opens, submission guidelines and the review rubric will be published here.