powerpoint-planning
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data to generate prompts for downstream AI tools, creating a potential surface for indirect prompt injection.
- Ingestion points: External data enters the workflow via 'Workflow 1: Brief' and 'Workflow 2: Source Handling', including reference materials and URLs used for slide content planning.
- Boundary markers: The skill provides mitigations such as the 'Source Classification' rules (Fact, Assumption, Hypothesis) and defined 'Human Review Triggers' for sensitive or customer-facing content.
- Capability inventory: The skill generates planning documents and production prompts for tools like Copilot; it does not directly execute code, perform network operations, or modify the system environment.
- Sanitization: There is no explicit instruction to sanitize or escape data before interpolation into the generated prompts, relying on manual gates for security.
Audit Metadata