agent-swarm-orchestration

Pass

Audited by Gen Agent Trust Hub on Sep 24, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill implements orchestration and routing patterns that process untrusted external data (such as user specifications and tasks) which can be used to influence agent behavior through indirect prompt injection.
  • Ingestion points: Untrusted data enters the agent context through the spec argument in Orchestrator.run_pipeline and the task argument in TaskRouter.route within SKILL.md.
  • Boundary markers: The implementation lacks explicit delimiters or instructions to the LLM to ignore potentially malicious instructions embedded within the user-provided data.
  • Capability inventory: The system is designed to execute actions via agent.execute(), potentially allowing for code generation and execution across multiple stages (PLANNER, CODER, TESTER) based on the input processing (SKILL.md).
  • Sanitization: The provided QualityGate template includes placeholders for checks like _check_no_secrets but does not implement logic to sanitize inputs or validate them against prompt injection patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 24, 2026, 06:27 PM
Security Audit — agent-trust-hub — agent-swarm-orchestration