Devin AI 6be376f804 Fix test failures: improve CustomLLM and error handling ...

- Fix CustomLLM to handle structured output for guardrails with JSON response
- Add proper method implementations (supports_function_calling, etc.)
- Handle 'Thought:' pattern like working CustomLLM implementation
- Change invalid LLM test to use LiteAgent instead of Agent
- Improve error messages to use type() instead of __class__
- Address GitHub review feedback for better error handling

Co-Authored-By: João <joao@crewai.com>

2025-07-07 18:37:03 +00:00

18 KiB

Raw Blame History

View Raw