dispatching-parallel-agents
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [SAFE] (SAFE): No malicious patterns, obfuscation, or data exfiltration attempts were detected. The skill is entirely documentation-based.
- [NO_CODE] (INFO): The skill consists of markdown instructions and conceptual pseudocode but lacks any executable scripts, configuration files, or automated logic.
- [Indirect Prompt Injection] (LOW): The skill describes a workflow for processing external code and test files. Evidence Chain: 1. Ingestion points: Test files and subsystems described in agent tasks. 2. Boundary markers: Absent in provided prompt templates. 3. Capability inventory: Implied file read/write and test execution permissions for dispatched agents. 4. Sanitization: Not addressed. The severity is low as the skill is purely educational and does not automate these operations.
Audit Metadata