recipe-update-doc

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill manages a workflow that processes external documentation content and user input, which presents a surface for indirect prompt injection.
  • Ingestion points: Reads existing files from project directories such as docs/design/, docs/prd/, and docs/adr/ in Step 1, and accepts user input via $ARGUMENTS and Step 3.
  • Boundary markers: Explicitly uses [STOP — BLOCKING] markers at critical stages (Steps 3, 5, and 6) to halt execution until the user provides confirmation.
  • Capability inventory: Spawns specialized sub-agents including technical-designer, prd-creator, code-verifier, document-reviewer, and design-sync to perform isolated tasks across SKILL.md (Step 4, 5, 6).
  • Sanitization: Employs a multi-layered verification strategy where updated content is reviewed by verification agents (code-verifier, document-reviewer) and requires final human approval before completion.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 10:14 PM
Security Audit — agent-trust-hub — recipe-update-doc