spoonos-testing-patterns

To mock LLM responses when testing SpoonOS agents, you apply standardized unit testing patterns that intercept and simulate model outputs. This allows you to validate agent logic and graph workflows deterministically without actual LLM API calls.

The best way to structure unit and integration tests for SpoonOS agents is by following a comprehensive testing pyramid. You can use pytest fixtures and structured templates to separate isolated logic tests from complex graph state and MCP tool integration tests.

Yes, these testing patterns support integration testing with MCP tools. You can validate SpoonOS agent interactions with external tools using structured test configurations and mocking utilities designed specifically for MCP workflows.

Yes, you can use pytest fixtures for SpoonOS agent graph and state testing. The patterns provide integration-ready templates that standardize graph workflow validation and state assertions within your pytest suite.

You configure CI-ready tests for SpoonOS agents by applying standardized pytest configurations and snapshot testing utilities. This approach ensures consistent test execution and debugging across automated development pipelines.