Agent Orchestrator

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONNO_CODE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The metadata includes an installation command that fetches the skill instructions from a remote GitHub repository using curl.
  • [PROMPT_INJECTION]: The skill defines orchestration workflows that are vulnerable to indirect prompt injection.
  • Ingestion points: The orchestrator processes user-provided task descriptions and requirements in the SKILL.md and through runtime interactions.
  • Boundary markers: The delegation templates and workflows lack explicit delimiters to separate instructions from untrusted external data.
  • Capability inventory: The agents being orchestrated have access to powerful tools, including GitHub for repository management, Supabase for database operations, and Playwright for web automation.
  • Sanitization: No validation or sanitization processes are defined for data passed between agents in the pipeline.
  • [NO_CODE]: The skill is composed entirely of markdown instructions and metadata without any bundled executable scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 02:42 PM