Plan de validation

Ce que l’Agent doit faire ensuite

1. 1Inspect repository, README/SKILL.md, license, and recent commits before production use.
2. 2Install in an isolated workspace or sandbox with no production secrets available.
3. 3Run the smallest representative task and record files touched, commands run, network access, and outputs.
4. 4Compare the selected skill against at least one alternative when the eval status is review or failed.
5. 5Promote only after the agent reports a successful verification result and unresolved warnings are accepted.

Ne pas utiliser quand

Conditions exigeant une autre skill

• Équipes qui nécessitent un SLA soutenu par le fournisseur
• production agents without a repository review
• Low GitHub adoption signal
• The SKILL.md excerpt is truncated; the full file must be complete and self-contained for the dispatcher to work correctly.
• The skill instructs reading files from the user's home directory (~/.everyday-writer/). While legitimate for voice profiles, this could be a privacy concern if the skill is used in a sandboxed environment; ensure the skill only accesses files the user has explicitly created and does not follow symlinks or read arbitrary paths.
• The sub-skills referenced in the dispatch map are not included in the submitted directory. The skill depends on them; ensure they are present in the repository and properly documented.