AI Agents From Scratch
ActiveDescription
Demystify AI agents by building them yourself. Local LLMs, no black boxes, real understanding of function calling, memory, and ReAct patterns.
Key Features
- 12-step progressive learning path from raw LLM calls to ReAct agents
- Runs entirely on local LLMs via node-llama-cpp, no API keys needed
- Covers function calling, memory, Tree of Thought, and error handling
- Companion website with visual explanations and concept maps
- Each example includes code, code explanation, and concept documentation
- Built-in benchmarks comparing local vs API model performance
Use Cases
💡 Learn how AI agents work under the hood before using production frameworks
💡 Teach students the fundamentals of agent architectures hands-on
💡 Prototype agent patterns locally without cloud dependencies or costs
💡 Understand ReAct, function calling, and memory systems from first principles
Categories
Quick Start
Clone the repo, run `npm install`, and download models into `./models/`. Start with `node intro/intro.js` and progress through examples sequentially. Requires Node.js 18+ and 8GB+ RAM. Visit agentsfromscratch.com for the companion guide.