problem-framing-canvas
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates a workshop by asking for user input and then restating that input in a refined format. This pattern creates a surface where a user could provide input designed to influence the agent's behavior, although no dangerous 'sinks' (like code execution or network access) are present in the skill.\n
- Ingestion points: User answers to prompts in
SKILL.md(Questions 1 through 6) are used to build the final canvas output.\n - Boundary markers: None identified; the agent is instructed to use the user's input directly to fill templates in Phase 3.\n
- Capability inventory: No file system access, network communication, or shell command execution is present in any of the skill's files.\n
- Sanitization: The skill performs no filtering or escaping of the user-provided text before it is interpolated into the final output.
Audit Metadata