Welcome to Dosu.
Dosu is the AI-powered knowledge base for engineering teams and the coding agents they work alongside. Dosu connects to the tools your organization already uses, turns the activity there into clear, up-to-date documentation, and serves that knowledge back to people and agents the moment they need it.
These docs explain what Dosu is, what it can do, and how to set it up. To get running fast, head to Quick Start. To see what happens behind the scenes, read How Dosu Works.
What Dosu does#
In a growing engineering organization, the knowledge that matters most lives in pull requests, chat threads, and half-written docs, and it goes stale the moment someone moves on. Dosu manages the full life of that knowledge so your teams and your agents can trust it:
- Connect your existing tools as sources. Dosu reads from GitHub, Slack, Notion, the web, and a growing list of integrations, so your knowledge base reflects how your organization actually works.
- Generate knowledge automatically. Dosu drafts documentation, changelogs, and answers from your code, issues, and conversations. Any answer Dosu produces can be saved as a lasting Document.
- Maintain it without the busywork. When a pull request merges, Dosu opens a pull request of its own with the doc updates. Monitors watch your sources and flag pages worth revisiting, so documentation keeps up with the code.
- Organize it into Libraries. Each Library is a knowledge base your teams and your agents can draw from, kept consistent with Topics and Templates.
- Put it to work everywhere. Dosu answers questions on GitHub and Slack and feeds project-specific context to coding agents like Claude Code via the Dosu MCP server and CLI.
Each of these maps to a section of these docs, so you can go as deep as you need.
Who Dosu is for#
Dosu is built for engineering organizations, and for the AI agents working alongside them. Platform and infrastructure teams use Dosu to give every engineer the same source of truth as the company grows. Teams use it to capture decisions and context before they are lost to turnover or buried in a channel no one searches. Coding agents use it to retrieve accurate, project-specific context instead of rereading the entire codebase on every run.
The same setup scales from a single repository to an entire company, and it works just as well for open source projects as it does behind the firewall.
You can get Dosu running in about five minutes and keep it in preview mode, reviewing responses privately before Dosu ever replies in public.
Key concepts at a glance#
You will see these terms throughout the docs. The Concepts section covers each one in depth, and the Glossary has quick definitions.
- Libraries are knowledge bases for projects, teams, or product areas. Sources, Documents, and settings live inside Libraries.
- Sources are the connected tools Dosu learns from, such as a GitHub repository or a Slack workspace.
- Agents are how Dosu shows up on a platform to answer questions and take action, for example on a repository or in a Slack channel.
- Monitors watch a source for changes worth documenting and flag knowledge that may be going stale.
- Topics group related documents so knowledge stays organized as it grows.
- Templates are reusable structures that shape how generated docs look.
- Review is where your team approves, edits, or declines Dosu's drafts before they go out.
Next steps#
- Quick Start walks you through connecting your first source and creating an Agent.
- How Dosu Works traces the path from a connected source to a published answer.
- Concepts explains each building block above in full.