workflow-automation
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features an AI Orchestrator pattern where external data is used to drive automated workflows.
- Ingestion points: In
references/detailed-guide.md, theaiOrchestratorfunction passesevent.data.taskdirectly into an LLM chat prompt to create an execution plan. - Boundary markers: The code examples do not demonstrate the use of delimiters or instructions to the LLM to ignore embedded commands within the task input.
- Capability inventory: The plan generated by the LLM leads to the execution of subtasks via
executeSubtask, which represents a wide capability surface for file, network, or system operations depending on the environment. - Sanitization: While the guide recommends validating plans against an allowlisted schema, the examples show direct interpolation of untrusted data without pre-processing or sanitization.
Audit Metadata