Llm Sandbox
Augments large language model inference with MCP client-server architecture for streamlined testing and agentic evaluation.
This sandbox provides a low-friction environment for validating Model Context Protocol (MCP) servers against a working LLM client using a live chat interface. Initially supporting local LLMs like LLaMA 7B for local network testing, it will expand to include cloud inference, enabling the use of more powerful models. It serves as both a reference architecture and a practical development environment, evolving alongside the MCP specification.
Key Features
01Local and cloud LLM inference support
02Facilitates agentic evaluation
03Live chat interface for interaction
04Reference architecture for MCP development
05MCP client-server architecture
061 GitHub stars
Use Cases
01MCP server validation
02Agentic workflow testing
03LLM behavior evaluation