parallel-agent
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of natural language instructions guiding an AI agent on how to manage parallel sub-tasks. It does not contain any executable code, shell commands, or network operations.
- [INDIRECT_PROMPT_INJECTION]: As an orchestration skill, it defines how the agent should handle user-provided tasks (ingestion points) using tools like
subagentandworkflowScript(capability inventory). The instructions include safety-conscious guidelines such as requiring 'parent-approved acceptedScope' and ensuring 'one repo/cwd or worktree has one writer' to prevent state corruption. While it processes external input, it establishes clear boundaries and approval workflows. - [REMOTE_CODE_EXECUTION]: There are no patterns of remote code execution. The skill mentions a
workflowScripttool, but it relies on an installed source of truth (pi-subagents) for the implementation and does not attempt to download or execute external scripts itself.
Audit Metadata