Progressive UI from LLM

📝 Markdown and HTML renderer for Svelte 5 — built for streaming AI agent output from Claude Code, ChatGPT, and agentic workflows. XSS-safe defaults, token caching, TypeScript types.

Incremental Markdown parser that emits streams of semantic events, plus tools to manipulate them — designed for real-time rendering of streamed LLM output.

Langchain Ollama Streaming example implemented in the flask

Streaming of LLM responses in realtime using Fastapi and Streamlit.

playgrounds for vercel ai sdk and langgraph, chat/streaming/resume...

Animated side-by-side demo of WebSocket vs Server-Sent Events. Same server, same LLM stream, every frame visualized in real time — with a kill/reconnect button to show how each transport recovers.

Zero-intrusion guard for LLM calls in dev: dedupe, cache, and protect AI requests across Node, browser, and Vite.

A high-performance, streaming-enabled Node.js bridge for the War-Machine local AI model. Optimized for i5-1235U CPU inference using Ollama and Express.

Custom Local AI Role-Play Framework

LLM-Talk enables natural voice conversations with language models using hotword activation.

MarkRender is a professional terminal markdown renderer designed specifically for streaming LLM (Large Language Model) responses. It renders markdown content directly in the terminal with rich formatting, syntax highlighting, and a flicker-free experience.

A FastAPI service that streams LLM responses token-by-token via Server-Sent Events, powered by a local Ollama model — built without LangChain to understand HTTP streaming at the protocol level.

implement llm streaming with page reload support using vercel ai sdk

A comparative engineering study benchmarking Next.js vs. Rails 7 for Clinical AI orchestration. Features edge-optimized RAG chains, streaming performance analysis, and architectural trade-offs in high-stakes CDS environments.

Go reference app for real-time LLM event streaming over SSE, backed by one Redis Stream with history replay and Last-Event-ID recovery.

Large Language Model (LLM) library for SWI-Prolog : Claude, ChatGPT, Gemini, Grok and Ollama. Supports streaming, sandboxed code execution and LLM to LLM communication

Type safe React hooks for real time WebSocket communication. Automatic reconnection, optimistic updates, and offline message queuing. Built for streaming LLM clients, real time dashboards, and collaborative apps.