claude-advanced-tool-use
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE] (SAFE): The skill is purely educational and provides integration templates for official AI features. It does not contain any malicious instructions or automated execution logic.
- [EXTERNAL_DOWNLOADS] (SAFE): The code snippets reference the 'anthropic' Python package, which is a trusted library for interacting with Claude models.
- [COMMAND_EXECUTION] (SAFE): The skill discusses 'Programmatic Tool Calling', but this refers to a platform-level sandboxed feature of the Claude API, not local command execution on the host system.
- [INDIRECT_PROMPT_INJECTION] (SAFE): Although the skill describes processing tool outputs, it includes a 'Guard + Act' pattern recommending server-side validation and security-first orchestration.
Audit Metadata