context-optimization Rapport d’évaluation.

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
• The SKILL.md excerpt is incomplete at the end, so the full masking rules and edge cases are not fully visible in the provided snippet.
• Indices de permissions à haut risque : Secrets or environment access
• Permission surface may require sandboxing
• The Python utilities use heuristic token estimation and simple summarization; the code itself notes that production systems should use model-specific tokenizers and LLM-based summarization.