claude-agent-sdk
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Mentions official installation of the Claude Agent SDK through standard package registries (pip and npm).
- [COMMAND_EXECUTION]: Describes the use of shell commands within the SDK framework for exploration and system operations.
- [DYNAMIC_EXECUTION]: Contains example code using
evalfor tool definitions, accompanied by explicit warnings to use secure alternatives in production. - [SAFE]: Includes a dedicated section on security best practices, covering input validation, permission callbacks, and secure credential handling.
Audit Metadata