What it is

Austen ships a Model Context Protocol (MCP) server, so the AI assistant you already use can run your content pipeline for you in plain language. Ask it to write an article, and instead of producing a generic draft, it hands the work to Austen. The result comes back in your brand voice, optimised for search and for AI answer engines, and saved to your account ready to edit and publish.

It is the same engine behind the rest of Austen, reached from wherever you work. You stay in the conversation. Austen does the research, planning, writing, and optimisation underneath.

Connecting is key-free

You do not generate a token, copy it, or paste it into a config file. Point your assistant at the Austen MCP endpoint and it walks you through a one-click approval in your browser. You sign in, review what you are granting, and click authorise. The assistant gets its own scoped permission and can keep working without you handling credentials. You can revoke it any time from Connected apps in your profile.

{
  "mcpServers": {
    "austen": {
      "url": "https://austenapp.com/mcp"
    }
  }
}

That is the whole setup. The first time the assistant uses Austen, it opens the approval screen for you.

What your assistant can do

Once connected, your assistant can:

Find your projects, so it writes for the right brand.
Draft a high-quality brief from a rough idea. Austen sharpens the angle, audience, and search intent before any writing starts, the same assessment used inside the app. You see the brief and can refine it.
Generate the full article from that brief, end to end: the plan, then the complete draft, on brand and optimised.
Track progress while the article is being written.
Return the finished article, with its metadata and SEO and GEO scores, ready to review.

A typical request looks like this:

"Use Austen to write an article for my blog about reducing churn for early-stage SaaS, targeting the keyword 'reduce churn'. Draft the brief first, show me, then generate it."

Your assistant drafts the brief, you approve or tweak it, and Austen produces the article. Because the work runs through Austen, the output sounds like you and is built to get found, which a one-off draft written in the chat window is not.

How it fits the rest of Austen

The MCP server is a front door to the same pipeline you use in the app. Briefs created through your assistant become real article briefs; generation builds a structured plan and then a full article, scored by SEO analysis and GEO analysis. Everything lands in your account, so you can open it, edit it, generate images, repurpose it, and publish, exactly as if you had started in Austen.

Prefer to build your own integration instead of using an assistant? The same pipeline is available over the developer API.

Who it's for

Anyone who lives in an AI assistant and wants real, on-brand content out of it rather than a throwaway draft. Founders who think out loud with an assistant and want the good ideas turned into published articles. Marketers who want to spin up a brief and a draft without switching tabs. Teams standardising on assistants who want those assistants to produce content that is actually on brand and search-ready.

Common questions

Do I need to paste an API key? No. Connecting uses a one-click browser approval. Your assistant gets its own scoped, revocable permission.

Which assistants work? Claude and any client that supports remote MCP servers. Clients that only speak stdio can bridge to the same endpoint.

Will it just write the article in the chat instead of using Austen? No. The server instructs connected assistants to route content requests through Austen rather than writing the draft themselves, so you get on-brand, optimised, saved output.

Where do the articles go? Into your Austen account, in the project you chose, ready to edit and publish like any other article.

Can I still use a token if I want one? Yes. A standard API bearer token works too, which is handy for headless setups.