icd10-cm-skill
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFE
Full Analysis
- Restricted Tool Operations: The skill is configured to interact exclusively with a specialized medical coding connector. It includes explicit instructions to cease operation if these specific tools are unavailable, which prevents the agent from attempting to use unverified external sources or memory-based recall.
- Minimal Capability Scope: The skill does not request or utilize capabilities for network communication, file system modifications, or shell command execution. By limiting its scope to specific data extraction and lookup tasks, it maintains a small security footprint.
- Data Processing Context: While the skill processes clinical documentation provided by users, its logic is focused on transforming that information into structured medical codes using authorized lookup tools within the platform's environment. There is no evidence of patterns that would lead to unauthorized data transmission.
- Instructional Integrity: The provided instructions are direct and purpose-built for the medical coding task. Analysis found no evidence of prompt injection techniques, obfuscated content, or attempts to override the underlying safety protocols of the AI agent.
Audit Metadata