The Agent Skills Directory

[SAFE]: The skill follows secure development practices by mandating a 'characterization-first' workflow, ensuring that automated tests document and verify existing behavior before any code modifications occur, thereby preventing regressions.
[SAFE]: Security is integrated into the hygiene process through a severity matrix that specifically prioritizes the identification and removal of hardcoded credentials and debug artifacts.
[PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection, which is an inherent risk for developer tools designed to process external or third-party codebases.
Ingestion points: Local codebase files are read and analyzed by parallel sub-agents during the 'SCAN' phase (Phase 2).
Boundary markers: The skill utilizes structured prompts defined in 'references/SMELL-DETECTION-PROMPTS.md' to guide sub-agent exploration, although these prompts do not include explicit instructions to disregard potentially malicious instructions embedded in the scanned files.
Capability inventory: The skill possesses the ability to modify the local filesystem (applying refactorings in Phase 5) and execute shell commands (running tests and linters in Phase 6).
Sanitization: This risk is effectively mitigated by a mandatory human-in-the-loop approval process for all significant refactorings and the requirement for fresh, automated verification evidence to confirm that behavior remains unchanged after every step.

supercharge