Using Pip Packages
LiteLambda supports almost any third-party Python library available on PyPI. We cache your dependencies to ensure blazing fast cold starts.
Adding Packages
To use external libraries, you must explicitly declare them in your job's configuration. In the dashboard, navigate to your Cron Job and open the Packages tab.
List your packages exactly as you would in a standard requirements.txt file. We highly recommend pinning exact versions to prevent your job from breaking unexpectedly due to a upstream library update.
# Good practice (pinned version)
requests==2.31.0
beautifulsoup4==4.12.2
# Acceptable, but risky for production
pandas
The Build Process
When you update your package list and click Save, LiteLambda kicks off an asynchronous background build process. We download the requested wheels, uncompress them, and store them in a dedicated layer cache for your specific job.
During execution, this pre-built layer is mounted directly into your sandbox container, allowing imports to resolve instantly.
Limitations
- Size Limit: Individual packages cannot exceed 10 MB compressed (wheel size). Packages like
tensorfloworpytorchare not supported. - C-Extensions: Packages that require complex system-level C dependencies during installation may fail to build. Standard pre-compiled manylinux wheels will work perfectly.