write-workflow-as-code

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to utilize the official wmill CLI for development workflows, including previewing local scripts, running deployed versions, and generating metadata files.
  • [REMOTE_CODE_EXECUTION]: The wmill script preview command allows for the execution of user-developed script code during testing. This is a core feature for the intended purpose of developing Windmill workflows.
  • [EXTERNAL_DOWNLOADS]: The wmill generate-metadata utility re-resolves and updates script dependencies from external package registries to ensure lockfiles are synchronized with the code.
  • [PROMPT_INJECTION]: The skill's functionality involves processing and executing code based on user instructions, which presents a surface for indirect prompt injection.
  • Ingestion points: The agent processes user-authored scripts and development requests to generate or modify workflow code.
  • Boundary markers: Standard script markers are used, but no specialized delimiters are specified for isolating code from instructions.
  • Capability inventory: The skill enables the agent to execute shell commands and run arbitrary script code via the wmill CLI.
  • Sanitization: Instructions do not include specific steps for the validation or sanitization of generated code prior to execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 04:44 AM
Security Audit — agent-trust-hub — write-workflow-as-code