The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (find-god-files.py and codebase-map.py) from the Arc plugin root to analyze codebase structure. These scripts are part of the skill's distribution and perform diagnostic tasks.- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external repository content and documentation. It manages potential injection risks by using specific safety prompts for its sub-agents and by producing plan-based outputs intended for human oversight. Ingestion points: Repository source code, CONTEXT.md, and ADRs; Boundary markers: Safety rules from references/subagent-safety.md are integrated into sub-agent prompts; Capability inventory: File system writes for RFCs and execution of local analysis scripts; Sanitization: The skill generates documentation for human review rather than executing code directly.- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were found. The skill operates within its defined scope of architectural analysis.

refactor