The Agent Skills Directory

[PROMPT_INJECTION]: The skill operates on external markdown files, creating an indirect prompt injection surface where malicious instructions in a target file could attempt to subvert the agent's optimization or verification processes.
Ingestion points: The content of the target SKILL.md file is ingested into the agent context in the main SKILL.md file and phases/01-analysis.md.
Boundary markers: The skill does not use explicit boundary markers or delimiters when reading the target file content.
Capability inventory: The skill has access to powerful tools including Bash, Write, Edit, Read, Glob, and Grep.
Sanitization: The skill lacks explicit sanitization of input data, relying instead on a structured inventory comparison in 'Phase 3: Integrity Check' to detect and revert unintended functional changes or losses.
[COMMAND_EXECUTION]: The skill interacts with the local environment using the maestro and ccw (Claude Code) CLI tools. These tools are used for context-aware operations such as loading coding specifications and searching project knowledge base entries, using controlled or hardcoded parameters.

skill-simplify