spec-driven-development
Pass
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill defines a structured development process (Spec-Driven Development) that mandates shared understanding and documented decisions through a transparent multi-phase workflow.\n- [PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface as it ingests raw requirements from users, but it includes strong architectural mitigations.\n
- Ingestion points: User requirements are recorded in Phase 1 within the
specs/*/instructions.mdfile.\n - Boundary markers: The skill utilizes Markdown headers and specialized templates to delimit user-provided content.\n
- Capability inventory: The workflow involves filesystem writes to a dedicated
specs/directory and command execution for running unit and integration tests viavitest.\n - Sanitization: The skill implements a robust manual sanitization strategy through six mandatory approval checkpoints; the agent is explicitly forbidden from proceeding between phases without human verification of the current output.\n- [COMMAND_EXECUTION]: The workflow involves running typechecks and behavioral tests (Phase 6). This is standard functionality for development tools and is governed by the 'Critical Rule' requiring explicit user authorization before execution.
Audit Metadata