spec-driven-development
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for indirect prompt injection because its primary function is to ingest user requirements and feature descriptions to generate executable specifications and code plans. However, the risk is minimized by design through an explicit gated workflow.
- Ingestion points: The skill processes user-supplied requirements, PRDs, or vague feature ideas provided in the initial agent turn.
- Boundary markers: The workflow mandates multiple human review checkpoints (Phase 0-4) where the user must approve the capability map, specification, plan, and tasks before implementation proceeds.
- Capability inventory: The skill defines the creation and execution of shell commands (
npm run build,npm test,npm run lint) and file system operations (writingSPEC-*.mdandtasks/*.mdfiles). - Sanitization: No automated sanitization is described, but the human-in-the-loop review process is the primary security control to prevent the agent from executing unintended or malicious instructions derived from external data.
Audit Metadata