solve-verification-challenge
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (such as image grids, text prompts, and audio clips from CAPTCHA providers) to determine browser actions. This represents an indirect prompt injection surface where a malicious site could attempt to influence agent behavior through challenge media.
- Ingestion points: Browser screenshots, OCR results, and audio transcripts derived from third-party verification widgets during the execution of the workflow defined in SKILL.md and references/solver-routing.md.
- Boundary markers: Absent. The instructions do not define specific delimiters or 'ignore' directives for content extracted from challenge media.
- Capability inventory: The skill utilizes browser control capabilities including clicking, dragging, typing, and invoking page-declared WebMCP tools as described in references/runtime-notes.md and references/webmcp.md.
- Sanitization: Absent. The workflow relies on raw perception (vision/audio) and transcription without filtering or escaping content derived from the untrusted challenge interface.
Audit Metadata