The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes benign discovery commands like pwd, find, and wc to map the codebase and determine the total file count. It also uses conditional shell logic to detect the technology stack based on specific manifest files.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface. Ingestion points: codebase files (.py, .ts, .js, .go, .rs, .java, .swift) processed from the directory structure. Boundary markers: not specified for the review subagents. Capability inventory: the skill utilizes the Task tool and several subagents (e.g., Explore, security-reviewer) to analyze code and implement fixes. Sanitization: no explicit sanitization or filtering of codebase content is mentioned.

hierarchical