evaluate-sdlc-layers 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 executes repository-specific commands (e.g., `uv run research/knowledge-explorer.py`) which could be risky if the repository is untrusted, but it is scoped to the user's own project.
• High-risk permission hints: Shell or command execution
• Permission surface may require sandboxing
• The `--fix` mode modifies files; while described as safe fixes, it could inadvertently alter files if the checklist logic is flawed.