SafeRL-Lab/cheetahclaws
A Python-based agent harness framework for building long-horizon, multi-model AI systems that integrate with Claude and support tool-use capabilities.
Velocity · 7d
+11
★ / day
Trend
→steady
star history
CheetahClaws provides infrastructure for developing and running autonomous AI agents capable of handling complex, multi-step tasks across extended time horizons. It supports multiple AI models, tool integration, and skill-based task execution. The framework includes security hardening, memory management, plugin/MCP sandboxing, and bridge integrations to platforms like QQ.