mb-from-prd
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided PRD text and '.memory-bank/prd.md' to decompose requirements into epics, features, and tasks. This creates a potential indirect prompt injection surface where malicious instructions embedded in a requirements document could influence the agent's planning and task generation process.
- Ingestion points: The skill reads '.memory-bank/prd.md' and processes user-supplied text inputs to define the project scope.
- Boundary markers: It implements checks for frontmatter status such as 'clarification_status: complete' and 'constitution_checked: true' and stops if 'NEEDS CLARIFICATION' markers are detected.
- Capability inventory: The skill performs file system write operations to create project documentation, epics, features, and task plans within the '.memory-bank/' directory, and it invokes project-specific commands like '/prd-to-tasks'.
- Sanitization: It relies on pre-processing steps like '/write-prd' and manual 'constitution' checks to verify the input data before processing.
Audit Metadata