form-1116
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted user tax data and 1099 forms to drive automated filing actions on official IRS sites.
- Ingestion points: Processes sensitive PII and foreign income facts (SKILL.md Prerequisites).
- Boundary markers: The instructions explicitly warn to "ask, don't guess" and halt for missing prerequisites.
- Capability inventory: Employs browser automation (Playwright/Puppeteer) for logging into and filing on the IRS website (filing.md).
- Sanitization: Includes a validation workflow (SKILL.md Step 9) requiring math and sanity checks before handoff.
Audit Metadata