The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to interact with Git for identifying review scopes and extracting code differences.
Evidence: git rev-parse, git log, git status, and git diff commands are utilized in both SKILL.md and the code-reviewer.md template.
[INDIRECT_PROMPT_INJECTION]: The subagent implementation ingests untrusted external data, such as code diffs and requirement documents, which could contain adversarial instructions intended to influence the reviewer's behavior.
Ingestion points: Variables like {WHAT_WAS_IMPLEMENTED}, {PLAN_OR_REQUIREMENTS}, and {EVIDENCE} in code-reviewer.md interpolate external content directly into the agent's prompt.
Boundary markers: The template does not use specific delimiters or instructions to prevent the agent from following commands embedded within the reviewed code.
Capability inventory: The agent has shell access to execute Git commands.
Sanitization: No explicit sanitization or validation logic is defined for the input variables.
[DYNAMIC_EXECUTION]: The code-reviewer.md template generates shell script blocks at runtime by interpolating variables into Git commands.
Evidence: The Git Review Scope section in code-reviewer.md constructs command lines using the {BASE_SHA} and {HEAD_SHA} placeholders for subsequent execution by the agent.

requesting-code-review