LangChain Agent Example

Install the real Python package:

python -m pip install 'keito @ https://github.com/osodevops/keito-python/archive/6da48aea9602e29e7a63dd85c588e05a508a25e2.tar.gz'
python -m pip install langchain-core langchain-anthropic

Create a configured LangChain Runnable in your application, then use the LangChain integration example with the local record_completed_work helper from Python Agent Integration. That helper is application code, not a Keito callback export.

The wrapper records one completed time entry after the invocation succeeds. It does not automatically infer token prices, create expenses, record every nested call or delete failed work. Use Expenses API for a separate explicit cost decision.