The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted data from GitHub pull requests, issues, and review comments in context-gathering.md. This data is used to inform the review and automated fix process.\n
Ingestion points: Pull request title/body, linked issue bodies/comments, and review thread comments are fetched using the gh CLI.\n
Boundary markers: The instructions do not define clear delimiters or warnings to the agent to disregard instructions contained within the ingested external content.\n
Capability inventory: The skill has access to powerful tools including Bash, Write, Edit, and Agent (for spawning subagents).\n
Sanitization: There is no evidence of sanitization or filtering of the fetched data before it is processed by the AI.\n- [COMMAND_EXECUTION]: The skill performs shell command execution to analyze code and verify fixes.\n
Tool usage: It utilizes standard development tools including git, gh, go, npm, cargo, and pytest.\n
Logic: It executes build, test, and lint commands specific to the project type (Go, Node.js, Rust, Python) to ensure that automated fixes do not break the codebase.\n
Argument handling: Parses user-supplied arguments in SKILL.md to determine the scope of the review.

review-deep