dispatching-parallel-agents

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOW
Full Analysis
  • PROMPT_INJECTION (SAFE): The skill contains instructional text and examples of how to draft focused prompts for sub-agents. These are standard task delegation examples and do not attempt to bypass safety filters or override the system prompt.
  • DATA_EXFILTRATION (SAFE): No network requests (curl, wget, fetch) or sensitive file path access are present. The file paths mentioned (e.g., src/agents/agent-tool-abort.test.ts) are illustrative examples for testing scenarios.
  • REMOTE_CODE_EXECUTION (SAFE): There are no remote downloads or executions. The code blocks provided are for documentation and conceptual purposes, illustrating how a user might invoke parallel tasks.
  • COMMAND_EXECUTION (SAFE): The skill does not contain any shell commands or subprocess calls. It describes a logical orchestration pattern rather than providing executable scripts.
  • OBFUSCATION (SAFE): The content is entirely human-readable markdown and dot-graph notation. No Base64, homoglyphs, or hidden characters were detected.
  • INDIRECT_PROMPT_INJECTION (SAFE): While the skill describes how to handle data from test failures, it is a static guide for the agent's reasoning process and does not create an automated pipeline for ingesting untrusted external data without supervision.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 08:02 AM