orchestration-patterns

Pass

Audited by Gen Agent Trust Hub on Mar 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists of documentation and code examples for legitimate agent orchestration patterns. No malicious instructions, obfuscation, or unauthorized data access patterns were found.\n- [PROMPT_INJECTION]: The orchestration patterns involve processing untrusted external data, which introduces a surface for indirect prompt injection.\n
  • Ingestion points: External source code (PR reviews), production logs (/var/log/production.log), and structured files (CSV, JSON).\n
  • Boundary markers: Not explicitly defined in example prompts; the skill suggests providing specific task contexts.\n
  • Capability inventory: Team and task management tools (TeamCreate, Task, TaskUpdate), internal messaging (SendMessage), and file access (Read).\n
  • Sanitization: The skill recommends pass-by-reference (using file paths) for large content to minimize context bloat and potential injection risks.\n- [COMMAND_EXECUTION]: The skill uses a framework-specific language for managing agents. All included commands (Task, SendMessage) are used within the intended scope of agent coordination and do not execute arbitrary system commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 18, 2026, 10:12 AM
Security Audit — agent-trust-hub — orchestration-patterns