The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by having a subagent process implemented code changes, which could contain adversarial prompts.\n- Ingestion points: The subagent reads implementation details and code diffs through placeholders in the code-reviewer.md template and via git diff and git show commands.\n- Boundary markers: The subagent prompt lacks explicit delimiters to isolate code content from instructions and does not include directives to ignore instructions embedded in data.\n- Capability inventory: The subagent is authorized to execute git shell commands, providing a potential path for exploitation if an injection succeeds.\n- Sanitization: No sanitization or validation is applied to the code content before it is processed by the subagent.

requesting-code-review