skills/security-phoenix-demo/security-skills-claude-code/phoenix-ambiguity-hunter/Gen Agent Trust Hub
phoenix-ambiguity-hunter
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown-based instructions and logic for an AI agent's persona. It does not include any scripts, binaries, or automated tool invocations.
- [SAFE]: No evidence of prompt injection, data exfiltration, obfuscation, or privilege escalation was detected. The instructions are aligned with the stated purpose of red-teaming security requirements.
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze external requirement specifications. This creates an attack surface where malicious instructions could be embedded within the requirements being reviewed. However, the skill lacks high-risk capabilities (like network access or arbitrary command execution) that would make such an injection dangerous.
- Ingestion points: External requirement documents or text input referenced in the skill's description (SKILL.md).
- Boundary markers: Not explicitly defined in the provided instructions.
- Capability inventory: The skill specifies output to a file (outputs/{session-id}/05-clarifications-v{N}.md), implying file-writing capability, but no other system-level tools are invoked.
- Sanitization: No explicit sanitization or filtering logic is defined for the input data.
Audit Metadata