safe-debug Eval report.

Validation plan

What the agent should do next

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.

Do not use when

Conditions that require another skill

• teams that need a vendor-supported SLA
• production agents without a repository review
• The skill references external shared principles files (../ai-research-reproduction/references/...) which may not be present in the skill directory, but this is acceptable as a shared reference and does not affect core functionality.
• No OpenAgentSkill engagement data yet
• High-risk permission hints: Shell or command execution
• The provided script excerpt is incomplete, but the full script is not required for evaluation; the skill's workflow is clear from SKILL.md.