workflow-orchestration
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The orchestration logic in
SKILL.mdcreates a surface for indirect prompt injection. - Ingestion points: Webhook triggers and input parameters entering the workflow context.
- Boundary markers: The
interpolatefunction performs direct string replacement without delimiters or security headers to distinguish data from instructions. - Capability inventory: The framework can execute LLM calls and MCP tools based on interpolated data, allowing malicious inputs to trigger actions.
- Sanitization: The code lacks validation or escaping for external content before use in prompts or tool arguments.
Audit Metadata