create-project-tasks
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a robust 'Grill the user' phase (Phase 2), which acts as a mandatory human-in-the-loop verification step. This ensures that the agent's understanding matches the user's intent before any write operations are performed in Notion.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Notion project pages and Figma design artifacts (Phase 1). This information informs the task creation process and codebase exploration.
- Ingestion points: External data is fetched from Notion URLs and Figma URLs provided as arguments or found on project pages.
- Boundary markers: The skill explicitly requires a multi-round 'interview' process to settle all design decisions before task generation.
- Capability inventory: The skill uses Notion and Figma MCP tools for reading/writing and an 'Explore agent' for local codebase read access.
- Sanitization: Verification is handled via the 'Grill the user' phase, which prevents automated execution of instructions embedded in external data. Because of these mitigations, this surface is considered safe.
Audit Metadata