workflow-greenfield

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill acts as a wrapper that passes user-provided input directly into a multi-step automated workflow engine.
  • Ingestion points: The user's prompt is ingested as the full context for the /workflow-start command in SKILL.md.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between the user's data and the workflow instructions, increasing the risk of indirect prompt injection.
  • Capability inventory: The resulting workflow has access to powerful capabilities including web research (/web-research, /deep-research), code generation and file system modifications (/scaffold, /cook), and test execution (/test, /integration-test).
  • Sanitization: No sanitization, validation, or escaping of the user context is performed before it is passed to the workflow sequence.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:25 AM
Security Audit — agent-trust-hub — workflow-greenfield