The Agent Skills Directory

[SAFE]: The skill implements a 'Plan Mode' which ensures that the agent remains in a read-only state while generating implementation plans. It explicitly prohibits editing code, writing data, creating branches, or performing any mutating operations during the planning phase.
[SAFE]: The instructions include a specific directive to avoid restating sensitive information such as raw logs, secrets, private payloads, or sensitive user data in the output, reducing the risk of accidental data exposure during the documentation and planning process.
[SAFE]: The skill's workflow is grounded in established engineering baselines and safety frameworks (such as OWASP LLM and NIST AI RMF), emphasizing the need for evidence-based reasoning and validation.
[SAFE]: No obfuscation, unauthorized remote code execution, or privilege escalation patterns were detected. External tool usage is limited to standard repository inspection tools (e.g., rg) as per the provided configuration.

fix-it