MCP App Store
by KeenEthics
Feedbaq icon

Feedbaq MCP Server for Claude and ChatGPT

By Peter Kildegaard

Overview

Feedbaq is the tool founders, product managers and marketers use to collect rich feedback from customers and build better products and campaigns. Describe what you want to learn, and Claude builds a short Feedbaq survey, publishes it when you say so, and gives you a link to share. Your customers answer out loud, or show you the problem on their screen while they talk. After each answer, an AI asks them one follow-up question about what they just said, so you get the reason behind the answer, not only the answer. When the responses come in, ask Claude what people said. It reads every transcript, finds the patterns, and quotes customers word for word, with a link to the exact moment in each recording. Use it for: - Customer discovery, without booking calls - Product reviews: what customers value and what gets in their way - Feature requests, with the problem behind each request - Churn: why people cancel, in their own words - In-the-moment feedback, right after someone uses a feature - Idea and concept tests before you build - Landing page, ad and message tests before you launch

How to connect

  1. Claude: Open the Feedbaq listing in the Claude connectors directory and choose Connect. Claude listing Feedbaq exposes 12 tools here.
  2. ChatGPT: Open the Feedbaq listing in the ChatGPT app directory and choose Connect. ChatGPT listing Feedbaq exposes 15 tools here.

Tools

create_study

ChatGPT
Create a new draft study with a welcome page, one empty voice question and a thank-you page. Read get_research_guide with the build stage (and the study type) first. The result includes the draft; pass the empty question's id to upsert_question as questionId to fill it. Use update_study for the pages.

delete_question

ChatGPT
Remove a question from a study draft. Published versions are not changed.

get_account

ChatGPT
Read the workspace this connection is bound to: its name, who the connection acts as (connectedAs: name and email), the connected user's role, whether it can publish (canPublish: an active plan or complimentary access), whether this user can subscribe, the plan terms, and the responses used in the current billing period. Call it once before you bring up publishing: when canPublish is true, publishing needs no talk about price.

get_research_guide

ChatGPT
Read Feedbaq's research method for one stage of a study. Call it before you start that stage. Stages, in order: onboarding (first conversation, no studies yet), discover (what to study), build (draft, preview and publish), share, analyze. For build, pass studyType to get the full definition of that study type: its questions, settings, follow-ups and rules. For analyze, pass studyType to get its analysis lens. Study types: churn-interview, concept-test, customer-discovery, feature-request, in-the-moment, landing-page, onboarding-check, pricing-reaction, product-market-fit, product-feedback, trial-conversion, usability-walkthrough.

get_response

ChatGPT
Read one complete response: every answer and follow-up, transcripts with word timings, and links into the app. Use it for the quotes in an analysis (see get_research_guide, analyze). questionUrl opens an answer; for a voice or screen answer, momentUrl opens the recording at a moment once {seconds} is replaced with a word's startMs / 1000. A choice answer names its option in value.optionId; a multiple-select answer lists every pick in value.optionIds. Pass questionId to read one answer only.

get_study

ChatGPT
Read a study: its draft, the list of published versions, and deep links. The draft is what upsert_question and update_study edit; the live version is the one participants see.

get_study_version

ChatGPT
Read the exact question wording from a published study version. Use a response's version_id to see the questions that participant answered.

invite_colleague

ChatGPT
Sends an invitation email to this address so they can join this workspace. Only workspace admins may do this.

list_responses

ChatGPT
List the responses to a study, newest first, a page at a time. Read get_research_guide with the analyze stage before you analyze them. Each response has its answers with transcripts and typed text, whether it counts toward the plan, and the published study version the participant answered. Word timings are left out; get_response has them. When nextOffset is not null, call again with that offset for the next page.

list_studies

ChatGPT
List studies in this Feedbaq workspace

publish_study

ChatGPT
Publish an immutable version of the current study draft, so participants can answer it. Call this only after the user has seen the preview and said a clear yes (the build guide from get_research_guide covers preview and publish). Publishing is the paid step: on a workspace without an active plan it fails and the error says what to do; get_account tells you in advance.

set_image

ChatGPT
Attach a public image (or video) by URL to a question, the welcome page, the thank-you page, or as the study logo. Pass mediaUrl null to remove it. Pages and the logo take images only.

update_study

ChatGPT
Rename a study, change its public address (slug) when the user asks for a custom one, set the brand name shown beside the logo, edit its welcome and thank-you pages, set the look of the participant page (background, accent, font), or set a response limit. Only the fields you pass change. The logo is set with set_image.

update_workspace

ChatGPT
Set up this Feedbaq workspace: the company or workspace name, the website, and the link address participants see (feedbaq.to/<linkAddress>/...), plus the connected user's own name. Propose the values first (read the website when you can) and save them only after the user says yes. Only workspace admins can change the workspace; anyone can change their own name. The link address cannot change once a study is published, because shared links would break.

upsert_question

ChatGPT
Add a question or update an existing question in a study draft. To change a question, pass its questionId and only the fields to change: everything you leave out keeps its current value, including edits the user made in the builder. Follow the build guide from get_research_guide: plain wording, one idea per question, behaviour before opinions, a follow-up instruction where the first answer is likely to be thin. Supports every question setting the builder has: text, choice, voice and screen-share formats, options with an Other choice, single or multiple select, skipping, alternative answer formats, AI follow-ups, screen task link and duration, phone and tablet behaviour, and position.

create_study

Claude

delete_question

Claude

get_response

Claude

get_study

Claude

get_study_version

Claude

invite_colleague

Claude

list_responses

Claude

list_studies

Claude

publish_study

Claude

set_image

Claude

update_study

Claude

upsert_question

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/feedbaq"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/feedbaq)