About validate
A repository-validation skill for maintainers of skill repositories. It runs freshness checks and explains how to address reported metadata, fixture, and stale-result issues.
Key features
- 01Run repository validation
- 02Check freshness of synced results
- 03Report metadata sync issues
- 04Identify missing evaluation fixtures
Use cases
- 01Validate a skill repository before publishing
- 02Find missing trigger evaluation fixtures
- 03Check whether results are stale after fixture changes
Install
Copy the command and run it in a terminal.
$
npx skills add https:/ / github.com/ datocms/ agent-skills --skill validateAfter installing, type/validatein Claude Code to confirm it loaded.
Details
From the project READMERun the repo validator with the freshness gate enabled:
python3 evals/scripts/validate_skill_repo.py --require-fresh-results-sync
If validation fails, read the error output carefully and fix each reported issue:
- Metadata sync errors: Update the
agents/openai.yamlfile to match the SKILL.md frontmatter for the affected skill. - Missing eval fixtures: Create
evals/fixtures/trigger/<skill-name>.jsonfor any skill that lacks one. - Stale results: a fixture changed after its results were committed. Do not re-run evals yourself: tell the user, who decides whether to re-run them or delete the stale result files.
After fixing, re-run validation to confirm all checks pass.
Comments
Markdown supported