The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process content from implementation files, story documents, and architecture records provided as input. \n- Ingestion points: Target implementation files, story files, and Architecture Decision Records (ADRs) are read using the Read tool throughout the review process (Phases 1, 3, and 7). \n- Boundary markers: There are no explicit delimiters or instructions provided to the specialist agents to ignore potential instructions embedded within the analyzed code or documentation. \n- Capability inventory: The skill utilizes Bash, Task, and Grep, providing a capability set that could be exploited if an agent follows malicious instructions hidden in the analyzed data. \n- Sanitization: No validation or sanitization is applied to the contents of the files being reviewed before they are passed to the specialists.

code-review