migrating-claude-agent-sdk-to-pydantic-ai
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a collection of Markdown instructions and reference tables designed to guide a migration process. It does not include any executable scripts or hidden code blocks.
- [CREDENTIALS_UNSAFE]: The skill does not contain hardcoded secrets or API keys. It correctly identifies that the application layer, not the agent SDK, should handle authentication and authorization.
- [EXTERNAL_DOWNLOADS]: All external references point to official documentation for Claude (Anthropic) and Pydantic. Mentioned dependencies such as
pydantic-aiandpydantic-ai-harnessare well-known libraries. - [COMMAND_EXECUTION]: The documentation discusses shell execution and coding tools in the context of migrating existing functionality. Crucially, it provides high-risk mitigation guidance, recommending the use of disposable workspaces, environment stripping, and cloud sandboxes (e.g., Modal) when dealing with untrusted code.
- [INDIRECT_PROMPT_INJECTION]: The skill guides the agent to read repository files to understand the source application. While this ingests untrusted data, the skill includes explicit high-risk gates and verification steps (e.g., using golden traces and deterministic models) to ensure parity and prevent unexpected behavior during the migration process.
Audit Metadata