Answer Box
Shubham Saboo's Awesome LLM Apps repository now offers over 100 hand-built, ready-to-run templates for AI agents, RAG pipelines, voice agents, and multi-agent teams, all under an Apache-2.0 license. The collection covers 15 categories including starter agents, advanced agents, always-on agents, MCP agents, and fine-tuning tutorials, with each template designed to run in three commands. The project is maintained by Shubham Saboo with community contributions and includes free step-by-step tutorials on Unwind AI.
A Cookbook for LLM Projects
Awesome LLM Apps is not a curated list of external links but a set of original, end-to-end tested templates that developers can fork, customize, and ship. The repository aims to eliminate the need to rebuild common patterns like RAG pipelines, agent loops, or MCP integrations from scratch. Each template is self-contained with full source code and runs with just three commands, avoiding broken dependencies or incomplete scaffolding.
The project supports multiple providers including Claude, Gemini, OpenAI, xAI, Qwen, and Llama, allowing users to switch between models with a configuration change. All templates are licensed under Apache-2.0, meaning they can be forked, shipped, or sold without paywalls, signups, or telemetry.
15 Categories of Ready-to-Run Templates
The repository is organized into 15 categories covering the modern AI stack. Starter AI Agents include single-file examples like an AI Travel Agent, AI Meme Generator, and AI Music Generator that run with just an API key. Advanced AI Agents feature production-style agents with tools, memory, and multi-step reasoning, such as an AI Deep Research Agent, AI Financial Coach Agent, and AI Investment Agent.
Multi-agent teams handle complex cross-domain tasks, with examples like an AI Competitor Intelligence Agent Team, AI Legal Agent Team, and AI Travel Planner Agent Team. Always-on agents run on schedules or events, monitoring context and proactively delivering updates. Voice AI Agents provide speech-in, speech-out capabilities using real-time voice APIs, including a Customer Support Voice Agent and Voice RAG Agent.
RAG, MCP, and Fine-Tuning
The RAG section includes 20 retrieval pipelines ranging from basic chains to agentic and multi-source systems, such as Agentic RAG with Reasoning, Hybrid Search RAG, and Knowledge Graph RAG with Citations. MCP AI Agents connect to external tools via the Model Context Protocol, with templates for Browser MCP, GitHub MCP, Notion MCP, and a Multi-MCP Agent Router.
Fine-tuning tutorials cover end-to-end recipes for open-source models, while LLM Optimization Tools help reduce token usage and API costs, including the Toonify Token Optimization and Headroom Context Optimization templates. The repository also includes crash courses on major agent frameworks like the OpenAI Agents SDK and LangGraph.
How to Get Started
To run the first agent, users can clone the repository and execute a travel agent template in three commands. The project encourages starring the repo to help other developers discover it and offers free step-by-step tutorials on Unwind AI. New templates and tutorials are delivered via email subscription.
The repository is created and maintained by Shubham Saboo with contributions from the community. It is licensed under Apache-2.0, allowing unrestricted use, modification, and distribution.