The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes instructions and content from external files (.claude/skills/*/SKILL.md, catalog.yaml, and quality-rubric.md) to perform its evaluations. This ingestion surface creates an opportunity for indirect prompt injection if those files contain malicious instructions intended to manipulate the agent's testing logic or assertions.
Ingestion points: The skill reads file content via Read, Glob, and Grep from the project's .claude/skills/ directory and the CCGS Skill Testing Framework/ folder.
Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are specified when interpolating the content of the skills being tested into the agent's context.
Capability inventory: The skill possesses the Write tool and is explicitly designed to update CCGS Skill Testing Framework/catalog.yaml and create result files in the results/ subdirectory.
Sanitization: There is no evidence of sanitization, filtering, or validation of the skill body text before it is evaluated by the agent.

skill-test