L

llm

A Zig implementation of the Model Context Protocol (MCP) spec revision 2025-03-26. Features JSON-RPC 2.0, stdio/HTTP transports, OAuth2.1 authentication, worker thread pool with cancellation, and deterministic testing.

Prompt Evaluator is a lightweight tool for evaluating and scoring AI-generated prompts and responses with F5 AI Guardrails. It helps assess quality, relevance, and safety using customizable criteria, making it ideal for testing LLM outputs in a structured, repeatable way.

High-level abstraction supporting code, data, and natural language description function definitions.