skill-brief
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill aggregates external content from prior art skills (sourced via find-skills) to generate a design brief for the skill-creator agent. This creates an indirect prompt injection surface where malicious instructions in external skill files could influence the behavior of the downstream agent.
- Ingestion points: The skill reads the landscape synthesis and candidate tables from find-skills, individual SKILL.md files from candidates, and descriptions of locally installed skills in ~/.claude/skills/.
- Boundary markers: The instructions do not mandate the use of delimiters (e.g., XML tags) or explicit instructions to ignore embedded commands when processing or passing external skill content to other agents.
- Capability inventory: The skill uses Read, Write, and Bash tools to manage scratch files in .claude/skill-brief/ and has the ability to invoke external skills like find-skills and skill-creator.
- Sanitization: There is no mention of sanitizing, filtering, or validating external skill content before it is interpolated into the capture brief or the handoff prompt.
Audit Metadata