The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands, specifically git diff, to retrieve code changes for the review process.
[PROMPT_INJECTION]: The skill ingests and processes untrusted data from the repository, including code diffs and design documents (e.g., NOTES.md, tech-spec.md). This creates an indirect prompt injection surface where instructions embedded in the analyzed content could attempt to influence the agent's review output or behavior.
Ingestion points: Data enters the context via git diff output and files located in the docs/ and .scratch/ directories.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the analyzed data are present.
Capability inventory: The skill includes file reading, command execution (git diff), and file writing to the .scratch/ directory.
Sanitization: No validation or sanitization of the analyzed content is described.

review