form-4562
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data that is interpolated into prompts and used within automated browser workflows, creating a surface for indirect prompt injection attacks.
- Ingestion points: User-provided asset lists, business names, and identifying numbers described in the Workflow and Prerequisites sections of
SKILL.md. - Boundary markers: The instructions do not define clear delimiters or include "ignore embedded instructions" warnings for the agent when handling user-provided strings.
- Capability inventory: The skill includes instructions for browser automation (Playwright, Puppeteer, Selenium, or Browserbase) to perform filing on external tax portals as described in
filing.md. - Sanitization: There are no explicit instructions to sanitize or escape user-provided text before it is used to fill form fields during the automated filing process.
Audit Metadata