anti-fabrication
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [No Code] (SAFE): The skill contains only markdown-based instructions and lacks any executable scripts, configuration files, or external dependency declarations.- [Safe Instruction Set] (SAFE): The content defines a rigorous framework for factual verification, reducing the risk of unintended behavior or fabricated information.- [Indirect Prompt Injection Surface] (SAFE): The skill establishes workflows that process data from the file system and command-line tool outputs. 1. Ingestion points: Tool outputs from Read, Grep, Glob, and Bash (SKILL.md). 2. Boundary markers: No explicit delimiters or instruction-ignore warnings are defined. 3. Capability inventory: Use of Bash, Read, Glob, and Grep tools (SKILL.md). 4. Sanitization: No explicit validation or escaping steps for tool outputs are mentioned.
Audit Metadata