Devin AI 142f3bbc60 This PR implements proper timeout handling for the max_execution_time parameter in the Agent class. When an LLM request times out, the system will now stop waiting for a response after the time defined by max_execution_time and either return a partial result or raise a TimeoutError. ...

Key changes:
- Added ThreadPoolExecutor-based timeout mechanism
- Implemented proper error handling and resource cleanup
- Added comprehensive tests for timeout functionality
- Ensured compatibility with existing agent execution patterns

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

2025-03-31 20:01:21 +00:00

60 KiB

Raw Blame History

View Raw