skill-to-workflow
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external agent instructions, prompts, and SOPs, which represents a surface for instructions embedded in source data to influence the agent. However, the skill instructions explicitly require the agent to avoid executing the business logic of the source material and to maintain a strict logical boundary between source semantics and the transformation task. 1. Ingestion points: Reads user-provided SKILL.md files, prompts, and related references. 2. Boundary markers: Explicit directives to 'not execute source skill business tasks' and to 'distinguish business semantics from execution mechanisms.' 3. Capability inventory: Limited to producing text-based documentation and specifications. 4. Sanitization: Employs structured output templates and evidence-based reasoning requirements to prevent instruction bleed.
- [SAFE]: The skill provides a structured methodology for technical documentation and workflow design without any dangerous operations. All referenced files are local and the logic focuses on analysis rather than system interaction.
Audit Metadata