dispatching-parallel-agents

Pass

Audited by Gen Agent Trust Hub on Aug 9, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill describes a pattern that creates an indirect prompt injection surface by delegating tasks to sub-agents using external input from test failures and system logs.
  • Ingestion points: Error messages, test failure logs, and subsystem descriptions provided by the environment or user are ingested and passed directly into sub-agent prompt structures (described in SKILL.md).
  • Boundary markers: The provided prompt templates do not include clear delimiters or explicit instructions for sub-agents to disregard instructions embedded within the ingested data.
  • Capability inventory: The agents receiving these prompts are granted broad capabilities to modify production source code and execute testing tools, which increases the potential impact of an injection attack.
  • Sanitization: There are no requirements or guidelines provided within the skill for escaping or validating the content of failure messages before they are used to generate new tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 9, 2026, 01:04 PM
Security Audit — agent-trust-hub — dispatching-parallel-agents