The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository-local data which could potentially contain malicious instructions.
Ingestion points: SKILL.md (Step 1) and claude-skill.md (Phase 2) read local instructions (AGENTS.md, CLAUDE.md), milestones, and design documents.
Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are present when reading these files.
Capability inventory: The skill possesses Read, Glob, Grep, and Write capabilities, and the instructions permit direct code modifications based on the gathered context.
Sanitization: The skill does not implement specific sanitization or validation logic for the content read from the repository files.
[SAFE]: The skill's primary function is to audit the codebase for production readiness, which includes identifying security risks like hardcoded credentials and debug output. All file operations and tool usages are consistent with this auditing and reporting purpose.

launch-checklist