spec-flow
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides a transparent and well-documented workflow for managing long-running tasks through structured documents and a local registry.
- [COMMAND_EXECUTION]: The workflow relies on a Node.js utility script located at .specs/bin/spec-flow.mjs. This script handles state transitions (init, new, checkpoint, accept, archive) by performing atomic JSON and filesystem operations within the workspace. It does not use unsafe execution functions like eval() or subprocess calls.
- [SAFE]: Data access is strictly scoped to the .specs/ directory in the current working directory. The skill does not access sensitive system paths, environment variables containing secrets, or hardcoded credentials.
- [PROMPT_INJECTION]: The instructional content focuses on the lifecycle actions and does not contain patterns designed to bypass safety filters or override system prompts. While the skill processes user-generated workspace data, it maintains a structured protocol with clear status tracking and verification steps that mitigate the risk of indirect instruction hijacking.
Audit Metadata