structured-output
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists of instructional text and Python templates using the Pydantic library for schema definition. No evidence of malicious behavior, exfiltration, or unauthorized command execution was found.
- [INDIRECT_PROMPT_INJECTION]: The skill documents how to process untrusted document text, which creates an attack surface for indirect prompt injection. 1. Ingestion points: The document string variable in the extract function within SKILL.md. 2. Boundary markers: Not explicitly defined in the provided conceptual prompt templates. 3. Capability inventory: No dangerous system capabilities such as subprocess calls, network access, or file system writes are present. 4. Sanitization: The skill recommends robust defensive sanitization including semantic validation and source-quote verification to ensure the model does not hallucinate or obey embedded instructions.
Audit Metadata