specialist-testing-testing-workflow-optimizer

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill defines a specialist agent for workflow optimization using standard instructional patterns. No malicious behavior, obfuscation, or unauthorized data access was identified.- [INDIRECT_PROMPT_INJECTION]: The skill acts on user-provided workflow data which is processed by a sub-agent. While this represents a potential surface for indirect injection, the implementation follows standard multi-agent patterns with a well-defined persona.
  • Ingestion points: User-provided business process descriptions and queries (SKILL.md).
  • Boundary markers: Absent; user input is interpreted within the context of the WorkflowOptimizer persona.
  • Capability inventory: Uses the Agent tool to spawn ephemeral sub-agents.
  • Sanitization: No explicit input sanitization is present, relying on the model's inherent safety guardrails.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 08:50 PM
Security Audit — agent-trust-hub — specialist-testing-testing-workflow-optimizer