wiki Informe de evaluación.

Plan de validación

Qué debe hacer el Agent después

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.

No usar cuando

Condiciones que requieren otra skill

• Equipos que necesitan un SLA con soporte del proveedor
• production agents without a repository review
• SKILL.md references several required reference files (operation-transactions.md, provenance.md, modes.md, plugins.md, rest-api.md) that are not present in the submitted file listing; if they are missing, the skill is incomplete for custom scaffolds and mutation workflows.
• Indicios de permisos de alto riesgo: ejecución de shell o comandos
• Permission surface may require sandboxing
• The core script is invoked from a hard-coded `/absolute/product/root/...` placeholder; the skill should explicitly instruct agents to derive the trusted product root from the skill installation location and verify it is not attacker-writable.