Teaching Guides

For contributors

This section explains Neo’s internals for people extending or contributing to it. If you just want to use Neo, you don’t need any of this — see Install and Quick start instead.

These guides explain Neo’s core features in plain language. The reference docs say what exists; these guides explain why it exists, how it works, and how to change it safely.

Guides

Guide	What it explains
Agent loop	How one user message becomes model calls, tool calls, and a final answer.
System prompt	What instructions Neo gives the model and how project context is added.
Tools	How Neo lets the model read files, search, run commands, and edit.
Permissions	Why approvals exist and how ask, trusted, and readonly work.
Providers	How Anthropic, OpenAI, OpenRouter, and Google Gemini plug in.
Sessions	How Neo saves and resumes conversations.
Compaction	Why long chats need context management and what Neo has today.

How To Read These

Each guide uses the same shape:

1. The simple idea.
2. The problem it solves.
3. How Neo implements it today.
4. How to customize or extend it.
5. What to be careful about.