The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill analyzes external code files which may contain adversarial instructions designed to influence the agent's behavior during the review process. 1. Ingestion points: Source code is read into the context using the Read, Grep, and Glob tools. 2. Boundary markers: No delimiters or explicit instructions to ignore embedded commands are present in the skill definition. 3. Capability inventory: The skill has access to the Bash and Task tools, which provide significant capability if the agent is manipulated. 4. Sanitization: No sanitization, escaping, or validation of the ingested code content is performed.

socratic-review