designing-data-intensive-applications 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
• No critical security risks found; the skill contains static reference material and does not instruct code execution, downloads, secret access, or destructive actions.
• SKILL.md is quite thin and mostly points to supporting files; the core workflow, expected output format, and limitations are not fully stated in SKILL.md itself.
• The relationship between the .mini.md, .md, and .nano.md files could be clearer, especially regarding when to escalate from one to another.
• The imperative 'OBEY' framing is unusual and could be mistaken for an injection-style instruction, though the content itself is benign.