agent-orchestration

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references official Google documentation and technical articles from trusted sources like Google Cloud and GitHub for educational purposes. These references are considered safe.\n- [PROMPT_INJECTION]: The described multi-agent patterns involve passing data between specialized agents, which is a standard architectural design. This creates a surface for indirect prompt injection if ingested data is not properly sanitized, though the skill itself is instructional.\n
  • Ingestion points: user_query passed to the coordinator agent and extracted data stored in context.state (SKILL.md).\n
  • Boundary markers: Not explicitly implemented in the high-level code templates.\n
  • Capability inventory: The agents are equipped with specialized tools like code_analyzer_tool, diagram_tool, and test_generator_tool (SKILL.md).\n
  • Sanitization: Not explicitly implemented in the provided architectural examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 09:08 AM
Security Audit — agent-trust-hub — agent-orchestration