hunt-dispatch
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's fingerprinting logic relies on strings extracted from external, untrusted sources such as HTTP headers, server banners, and HTML content to dynamically load specialized security skills. This creates a surface where an attacker-controlled environment can influence the agent's choice of tools and focus.
- Ingestion points: The skill parses Location headers, version-specific strings in HTML, framework markers like NEXT_DATA, and service banners such as okta.com or MongoDB.
- Boundary markers: The instructions mandate a confirmation gate requiring user authorization and explicit scope definition before any active probing or exploitation.
- Capability inventory: The skill is designed to load modules for remote code execution (RCE), SQL injection, SSRF, and credential bypass, signifying a high-impact capability tier.
- Sanitization: While it defines a precedence order for loading skills, it lacks specific instructions for sanitizing or escaping the data strings used for fingerprinting.
Audit Metadata