utility-pm-workflow-builder

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting user-provided workflow ideas and evidence and interpolating them into generated markdown drafts.
  • Ingestion points: User input in Step 1 (Workflow Idea) and Step 2 (Why Gate evidence).
  • Boundary markers: The generated drafts lack explicit delimiters or instructions to ignore potential commands within the user-supplied text during interpolation into the PACKET.md and workflow.md drafts.
  • Capability inventory: The skill instructions involve file system reads (scanning _workflows/ and skills/ directories) and file writing to a specific gitignored staging path (_staging/workflows/). It does not have network access or arbitrary command execution capabilities.
  • Sanitization: There is no evidence of input validation or escaping for the user-supplied strings before they are written to the draft files.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 03:11 PM
Security Audit — agent-trust-hub — utility-pm-workflow-builder