llm-as-judge
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious code, command execution, or network exfiltration patterns were detected. The skill is entirely composed of instructional markdown without any functional scripts or tool invocations.
- [INDIRECT_PROMPT_INJECTION]: The skill outlines a workflow for processing untrusted candidate text, establishing an attack surface. Mitigation is provided within the instructions. 1. Ingestion points: Untrusted candidate response text (SKILL.md). 2. Boundary markers: Instructions mandate the use of explicitly delimited data blocks and 'ignore instructions' warnings (SKILL.md). 3. Capability inventory: The skill does not define any tools, shell commands, or network operations. 4. Sanitization: Recommends structural firewalls and content delimiters to separate data from instructions.
Audit Metadata