# basemode

> Documentation for basemode, the context engine that briefs your coding agents from a graph of your workspace.

## Docs

- [Getting started](https://docs.basemode.ai/index.md): Install base, wire it into Claude Code, and teach it the first thing it will tell you back. By the end of this page a rule you wrote by hand arrives inside an agent turn without anyone asking for it.
- [How it works](https://docs.basemode.ai/how-it-works.md): There are four points inside a single agent turn where base can place context, and it uses all of them. This page covers what gets placed at each one, how it decides what is relevant, and what base is holding that it can draw on.
- [Command reference](https://docs.basemode.ai/reference/commands.md): Every base command grouped by consequence: the ones that are safe to run at any time, the ones that write to the graph, and the ones that remove things. Generated from the shipped 0.13.2 binary, not from a hand-maintained list.
- [Changelog](https://docs.basemode.ai/changelog.md): Released versions of base, newest first. Each entry covers what changed in the command surface and anything that affects an existing install.