The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes external content from code comments, which constitutes an indirect prompt injection surface. Malicious instructions could theoretically be embedded in these comments to mislead the agent. However, the skill's instructions are focused on pattern detection and rewriting rather than execution, and it requires user validation for fixes. * Ingestion points: Source files and documentation (.go, .py, .js, .ts, .md, .txt) read by the agent. * Boundary markers: None identified. * Capability inventory: Access to Read, Write, Bash, Grep, Glob, Edit, and Task tools. * Sanitization: None provided for comment contents.
[COMMAND_EXECUTION]: The skill uses Bash to execute grep and rg commands for pattern identification. These operations are limited to searching for specific keywords and do not involve executing untrusted input.

comment-quality