dual-steelman
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a set of text-based instructions designed to guide the agent through a logical reasoning framework (Steelmans). It does not request or use any dangerous tools or system permissions.
- [EXTERNAL_DOWNLOADS]: The
SKILL.mdfile contains a URL (mp.weixin.qq.com/s/6eDElggMR7aefaEzlWfVMA) in the 'Sources' section to attribute the methodology to its original author. This is a reference to a well-known content platform and is used for documentation purposes rather than runtime code execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate its analysis. While it lacks explicit boundary markers or sanitization, it does not possess any dangerous capabilities (e.g., file writing, network requests, or tool execution) that could be exploited through injection. The structured nature of the response protocol (requiring a 'crux' and 'steelman' arguments) naturally enforces a level of critical processing on the input.
- Ingestion points: User-provided decisions or stances described in
SKILL.mdinstructions. - Boundary markers: None identified.
- Capability inventory: Limited to text generation; no subprocess calls, file operations, or network tools are enabled.
- Sanitization: None identified.
Audit Metadata