howl-anderson/agentsilex
A minimal, transparent agent framework in ~300 lines of Python code for developers who want full control over their AI agents.

Velocity · 7d
+1.1
★ / day
Trend
→steady
star history
AgentSilex is a lightweight agent framework designed for transparency and hackability. It provides a clean, readable codebase that developers can understand, customize, and extend. Built on LiteLLM, it supports 100+ LLMs including OpenAI, Anthropic, Google Gemini, and local models. The framework is suitable for companies building custom agent systems, developers learning agent internals, and educators teaching AI concepts.