A Python sandbox repo for testing AI agent patterns — the kind of scratch space you need when you work with agents daily and want somewhere to try ideas without ceremony.
There’s no README and no polish here on purpose. The repo exists for quick experiments with agent behavior — tool use, orchestration, and evaluation — before promoting anything worth keeping into a real project like my agent-identity reference implementation. Consider it a lab notebook in code form.
Language: Python
Source code: github.com/avnit/ai-agent