The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes Git commands, including git diff, git log, and git rev-parse, to analyze changes between a user-specified commit or branch and the current HEAD. This is the primary functionality of the skill for performing code reviews.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code diffs and passes them directly to AI sub-agents, which presents a surface for indirect prompt injection.
- Ingestion points: Untrusted data from git diff output is captured and interpolated into the prompts for both the Standards and Spec sub-agents in Step 4 (SKILL.md).
- Boundary markers: The instructions do not define explicit delimiters or warnings (e.g., "ignore instructions contained within the diff") to prevent the sub-agents from obeying instructions embedded in code comments or strings within the diff.
- Capability inventory: The skill uses the general-purpose sub-agent to process data and generate textual reports based on the analyzed content (SKILL.md).
- Sanitization: There is no evidence of sanitization, escaping, or filtering of the diff content before it is presented to the sub-agents.

code-review