hunt-cache-poison
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONOBFUSCATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to interact with untrusted external web servers and process their responses (headers and body content), which creates a potential surface for indirect prompt injection.\n
- Ingestion points: Target server responses during discovery and validation phases (e.g., mapping cache infrastructure, checking header reflection in SKILL.md methodology).\n
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' markers for the content fetched from target servers.\n
- Capability inventory: The agent is expected to use network tools (curl) to interact with targets as described in the 'Payload & Detection Patterns' section.\n
- Sanitization: The skill lacks instructions for sanitizing or escaping content received from target servers before the agent evaluates it against expected patterns.\n- [OBFUSCATION]: The skill uses URL encoding within its payloads to demonstrate security bypass techniques.\n
- Evidence: The string 'evil%2ecom' is used in the 'Bypass Techniques' section to illustrate how to evade WAFs filtering 'evil.com'.
Audit Metadata