autodev-parallel
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill functions as a productivity tool for orchestrating sub-agents in a controlled git environment.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by processing external PRD files.
- Ingestion points: Reads task descriptions from PRD.md in Phase 1.
- Boundary markers: Absent; instructions are directly interpolated into sub-agent prompts.
- Capability inventory: Employs Bash, Write, Edit, and Agent tools to modify code and execute commands.
- Sanitization: Absent; the skill trusts the content of the PRD for task definition.
- [COMMAND_EXECUTION]: Executes the verify command specified in the configuration to validate changes. This is a standard part of the automated implementation loop.
- [DATA_EXPOSURE]: Accesses local configuration files for team roles located at ~/.claude/team-roles/. While these are in the home directory, they are limited to the agent's own metadata and persona settings.
Audit Metadata