autodev-parallel

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes user-provided strings through the Bash tool, specifically for the verify command and within the sub-agent coordination loop.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads and interprets the contents of PRD.md or other specified requirement files to drive the autonomous development loop, creating a surface for data-driven instruction injection.
  • Ingestion points: Project requirement files (e.g., PRD.md) and the standard output of sub-agents (AUTODEV_RESULT).
  • Boundary markers: Sub-agent prompts use triple-quoted blocks, but there is no explicit logic to prevent the orchestrator from following instructions embedded within the PRD items themselves.
  • Capability inventory: Access to Bash, Write, Edit, Read, and Grep. The skill can execute shell commands, modify source code, and perform git operations.
  • Sanitization: No sanitization or validation is applied to the requirements text before it is interpolated into sub-agent prompts or used for logic branching.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 01:41 PM
Security Audit — agent-trust-hub — autodev-parallel