On January 9, 2026, the AI Trading project encountered an issue during the removal of LangSmith from its codebase, highlighting the challenges of maintaining code hygiene in dynamic development environments. The incomplete removal left behind try/except import blocks in four source files, potentially causing ImportError if those code paths were executed. This incident underscores the importance of thorough code cleanup procedures when removing dependencies.

The project's developer failed to update all files that imported from the LangSmith module, a common oversight in dependency management. To address this, the team replaced all try/except import blocks with a flag, LANGSMITH_AVAILABLE = False, ensuring that the code gracefully handles the absence of LangSmith operations. This lesson emphasizes the necessity of a systematic approach to code cleanup, involving steps such as running grep commands to list all files that import from a module, updating all importing files before deletion, and running a full test suite to verify the integrity of the codebase.

The incident also highlights the shift towards Vertex AI RAG as a replacement for LangSmith observability, reflecting ongoing advancements in AI trading infrastructure. This experience serves as a valuable lesson for developers and teams involved in managing complex codebases, emphasizing the need for rigorous processes in dead code removal to avoid potential runtime errors and maintain code quality.