agent-workflow-designer
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's analysis tools ingest user-provided JSON files for workflow validation and cost estimation. \n
- Ingestion points: The scripts
workflow_validator.py,cost_estimator.py, andworkflow_visualizer.pyread JSON data from file paths or standard input. \n - Boundary markers: While the tools operate on structured JSON, the instructions do not specify prompt delimiters to isolate these inputs when discussed by the agent. \n
- Capability inventory: The provided scripts are limited to local data processing and CLI output; they do not have network or file-system write permissions. \n
- Sanitization:
workflow_visualizer.pyimplements ID sanitization to prevent injection into Mermaid diagram syntax.\n- [SAFE]: No malicious patterns, obfuscation, or unauthorized capability requests were identified. The skill author 'borghei' provides legitimate utility scripts and documentation consistent with the skill's purpose.
Audit Metadata