qa-agent-claude

Pass

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an inherent surface for indirect prompt injection due to its automated QA workflow.
  • Ingestion points: In SKILL.md, Step 1 — Explore instructs the agent to crawl a provided application entry point to record routes, forms, and interactive elements.
  • Boundary markers: The instructions do not provide explicit delimiters or warnings to the agent to treat crawled content as untrusted data or to ignore embedded instructions found within the application's HTML or metadata.
  • Capability inventory: In SKILL.md, Step 3 — Generate and Step 4 — Run provide the agent with the capabilities to write code to the local repository and execute it using browser automation tools.
  • Sanitization: The skill lacks instructions for sanitizing or validating the structural data discovered during exploration before it is used to derive user journeys and generate executable code.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 8, 2026, 02:32 PM
Security Audit — agent-trust-hub — qa-agent-claude