anti-sycophancy
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to process and extract claims from user-provided input, which creates a surface for indirect prompt injection. However, the skill does not use any high-risk tools (network, file system, or subprocesses), and the risk is inherent to the skill's primary purpose of analyzing user input to provide objective responses.
- Ingestion points: User messages containing claims and framing.
- Boundary markers: None explicitly defined in the instruction set.
- Capability inventory: No dangerous capabilities (file-write, network, execution) are defined in the skill.
- Sanitization: No specific sanitization of the extracted claim is performed.
Audit Metadata