smolagents

smolagents belongs in the database because it is one of the clearest examples of a lightweight, code-first agent framework that does not try to become an entire AI platform. The official Hugging Face docs describe it as an open-source Python library designed to make building and running agents extremely easy in just a few lines of code. The docs also highlight its two main paradigms: CodeAgent and ToolCallingAgent. That matters because smolagents has a very specific appeal to builders who want agent abstractions kept small and understandable.

It is also worth including because the economics are simple and honest. The library itself is open source and free, while costs depend on the model providers and execution environments used underneath it. The project also references sandboxed execution via Modal, Docker, E2B, and others, which reinforces that smolagents is a framework layer rather than a hosted SaaS. For the database, it should be positioned as a minimal agent-building framework for technical teams, not as a commercial platform.