Agent-native documentation

Docs that never fall out of date.

Connect your GitHub repo and plumedoc generates a beautiful, always-current documentation site straight from your code — and publishes it into a network that AI agents can discover, search, and trust.

Reference-first · Beautiful by default · llms.txt + MCP built in

Why plumedoc

Documentation that keeps up with your code — and your readers.

Existing tools solve authoring and hosting. plumedoc treats “always accurate from source” and “agent-native distribution” as the whole point.

Always current

Docs rot the moment code ships. plumedoc regenerates reference straight from your source and keeps it tracking main — so what readers see is what you actually shipped.

Beautiful by default

Great docs are a design project most teams never staff. plumedoc ships a site you're proud of out of the box — reference-first, typographically careful, light and dark.

Agent-native

AI agents are first-class readers now. Every site exposes clean markdown, an llms.txt map, and an MCP endpoint — so agents can find, parse, and trust your product.

The differentiator

Built for the agents reading your docs — not just the humans.

Developers increasingly reach for a product through an agent. If an agent can't find, parse, and trust your docs, your product is invisible to a fast-growing slice of your users.

  • llms.txt, by default

    Every site publishes llms.txt and llms-full.txt — a clean, stable map an agent can pull in one request, no scraping.

  • A per-workspace MCP endpoint

    Agents query a single product's reference on demand over MCP — the same source of truth your docs render from.

  • The agent-queryable network

    Each site joins a shared, searchable index. Agents discover products across it — and being absent becomes the disadvantage.

llms.txtGET /acme-sdk/llms.txt
# Acme SDK

> Type-safe client for the Acme API.

## Getting started
- [Quickstart](https://acme-sdk.plumedoc.com/quickstart)
- [Installation](https://acme-sdk.plumedoc.com/install)

## Reference
- [Client](https://acme-sdk.plumedoc.com/reference/client)
- [Errors](https://acme-sdk.plumedoc.com/reference/errors)
How it works

From repository to a living docs site in three steps.

  1. 01

    Connect your repo

    Install the plumedoc GitHub App and point it at a repository. Compose multiple repos into one coherent site as you grow.

  2. 02

    Agents generate the docs

    A harness reads your codebase and writes reference-first documentation — API surface, types, endpoints, and examples pulled from real code.

  3. 03

    Published & kept current

    Your site ships beautiful by default and regenerates as you push, so what readers and agents see always matches main.

Give your product docs agents can trust.

Connect a repo and get a beautiful, always-current site in minutes — on the network the models will learn your product from.