BigHugger
GH Repository · lsdefine

GenericAgent

Self-evolving agent: grows skill tree from 3.3K-line seed, achieving full system control with 6x less token consumption

stars
14,190
30-day movement
starts with the next reading
Related entries
60
Connections
1
agent-frameworktask-automationskill-treeai-agentllm-agentautomationautonomous-agentbrowser-automationclaudecomputer-controlgeminidesktop-automationPythonmemory-systemlightweightself-evolvingpython

GenericAgent is a Python agent framework that starts from a small (~3.3K-line) seed codebase and self-evolves by growing a skill tree of learned capabilities. It targets full computer control—desktop and browser automation—and reports using roughly 6x fewer tokens than comparable approaches.

Reach for it when you want a lightweight, self-improving agent that controls your computer while keeping token costs low.

Use it to

  • Automate desktop tasks
  • Control the browser via agent
  • Grow a reusable skill tree
  • Run tasks with lower token spend
  • Build on a small seed codebase

For Python developers building autonomous computer-control agents

Role
agent-framework
Language
Python
Licence
MIT
Forks
1,651
Open issues
131
Last push
2026-09-14
Latest release
v0.1.0 · 2026-05-15
topicsagent-frameworkself-evolvingcomputer-controlautomationllm-agentpython