Ask your reader feedback anything: the ReaderPulse MCP server

Your beta readers leave a trail of reactions, comments, and ratings across every chapter you share. Until now, the only way to explore that feedback was to open ReaderPulse and read through it pane by pane. Today we’re changing that.

ReaderPulse now ships an MCP server — a secure connection that lets an AI assistant like Claude read your reader feedback directly, with your permission. You can ask questions about your manuscripts in plain language and get answers grounded in what your real readers actually felt.

What is MCP, in plain terms?

MCP (the Model Context Protocol) is an open standard for connecting AI assistants to the tools and data you already use. Think of it like a secure doorway: instead of copying and pasting feedback into a chat window, you grant your assistant a scoped, read-only key to your ReaderPulse account. From then on it can look up your manuscripts, chapters, reactions, and ratings on demand — and nothing else.

You stay in control the whole time. Access is granted through a normal login-and-approve screen, it’s strictly read-only, and it only ever sees the manuscripts that belong to you.

What you can ask

Once connected, your assistant can pull the same insights you see in the dashboard. A few examples of what becomes possible:

• “Which chapter in my latest draft made readers cry the most?” — it reads your per-chapter reaction breakdown and points you to the moment that landed.
• “Summarize the written feedback on chapter 3.” — it gathers every comment readers left on that chapter and distills the themes.
• “Where did readers get confused or bored?” — it surfaces the exact highlighted passages tied to those reactions, so you know what to revise.
• “What’s my average rating, and how has it changed between versions?” — it compares story-level ratings across drafts.
• “Catch me up on the latest reader activity.” — it reads your recent activity feed across every manuscript.

Under the hood, the server exposes a focused set of read-only tools: listing your manuscripts and their versions, per-chapter analytics, inline reactions with the passages they’re attached to, chapter-level and story-level written feedback, and your recent activity feed. Your assistant chooses which to call based on what you ask.

How to connect it

Setup takes a couple of minutes and only has to happen once. Using Claude as an example:

1. In Claude’s settings, open Connectors and choose to add a custom connector.
2. Paste in the ReaderPulse server URL:

   https://www.readerpulse.io/api/mcp

3. You’ll be sent to a ReaderPulse sign-in screen. Log in with your normal account, then click Allow to approve access.
4. That’s it. Head back to your assistant and start asking questions about your manuscripts.

The connection uses the same secure OAuth flow you’d expect from any trusted app, so you never share your password with the assistant and you can revoke access at any time. Because every request is scoped to your account, your assistant can only ever read your own manuscripts and the feedback on them.

Why we built it

Reader feedback is most useful when it’s easy to think with. The MCP server turns your ReaderPulse data into something you can interrogate conversationally — spotting patterns, comparing drafts, and deciding what to revise next, without manually combing through every comment. It’s the same honest reader insight you already trust, now available wherever you do your thinking.

Don’t have feedback flowing in yet? Create a free account and share your first manuscript — then connect your assistant and let it help you make sense of what your readers feel.