first-principles
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text through a placeholder in its reasoning prompts. While this is an entry point for external content, the skill lacks any dangerous capabilities such as tool access, file system writes, or network operations, meaning any potential injection is confined to the text generation context.
- Ingestion points: User-supplied problem descriptions provided to the 'Decompose' prompt in SKILL.md.
- Boundary markers: Not explicitly defined in the prompt structure.
- Capability inventory: None. The skill does not use any tools or execute code.
- Sanitization: None.
Audit Metadata