The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by instructing a subagent to analyze code changes produced by git commands.
Ingestion points: The code-reviewer.md template directs the subagent to process the output of git diff {BASE_SHA}..{HEAD_SHA} and implementation descriptions.
Boundary markers: The instructions lack clear delimiters or specific prompts to ignore potential instructions embedded within the source code being reviewed.
Capability inventory: The subagent is empowered to provide production readiness assessments and merge verdicts, which could be manipulated by adversarial content in the diff.
Sanitization: There is no evidence of sanitization or filtering of the code content before it is processed by the LLM.
[COMMAND_EXECUTION]: The skill utilizes standard version control commands to identify changes.
Evidence: SKILL.md and code-reviewer.md use git rev-parse, git log, and git diff to determine the scope of the review. These commands are typical for the skill's stated purpose of facilitating code reviews within a development environment.

requesting-code-review