parallel-agents
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFE
Full Analysis
- [SAFE] (SAFE): The skill is purely instructional, providing templates and patterns for multi-agent orchestration within Claude Code.
- [SAFE] (SAFE): Requested tools (
Read,Glob,Grep) are standard search and read utilities for codebase navigation and do not involve system modification or network access. - [SAFE] (SAFE): No external dependencies, remote script downloads, or package installations are present.
- [SAFE] (SAFE): No evidence of obfuscation, hardcoded credentials, or data exfiltration patterns.
- [Indirect Prompt Injection] (SAFE): While the skill defines agents that read codebase data (an ingestion surface), it does not implement unsafe operations like
eval()or network sends that would create a high-risk vulnerability chain. The use-case is standard for AI-assisted development.
Audit Metadata