ai-python-custom-loop
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- Standard Agent Loop Template: The skill provides structural guidance for managing tool dispatch and control flow in an agent loop. It uses established SDK patterns that do not introduce unusual risks.
- Security Best Practices for Tool Execution: The instructions emphasize using
context.resolve(event.tool_call), which is noted to handle validation and approval gates. This is a positive security pattern that helps prevent unauthorized tool execution. - Indirect Prompt Injection Surface: Like any agent loop that processes message history, the implementation is subject to indirect prompt injection if untrusted data is processed. The skill mitigates this by instructing developers to use built-in validation mechanisms.
- Vendor-Owned References: The skill references internal concepts like
ai-python-serverless-executionandai-python-durable-execution, which are consistent with the infrastructure provided by the vendor for managing execution boundaries and durability.
Audit Metadata