Pheedback MCP Server
Connect Claude and ChatGPT directly to your Pheedback workspace. Create surveys, list experiences, and pull response data through natural language.
Overview
The Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools and read live data. The Pheedback MCP server exposes your workspace over this protocol, so you can ask an AI to "create a survey about onboarding friction" or "show me the last 20 responses from the NPS pheedback" without leaving your AI client.
The server provides four tools:
list_organizations— discover which organizations your connected account has access tolist_pheedbacks— browse surveys/experiences in an organizationcreate_pheedback— create a new survey from a name and goalget_pheedback_responses— retrieve visitor responses with pagination
Setup
Pheedback connects to Claude and ChatGPT through their connector marketplaces — no manual config files or API keys required. The connection is authorized via OAuth using your existing Pheedback account.
Claude Desktop
Search for Pheedback in the integrations marketplace and click Connect. You'll be redirected to Pheedback to authorize access, then returned to Claude with the tools ready to use.
Claude Code (CLI)
Add the server to your project or global settings:
claude mcp add --transport http \
--callback-port 5432 \
Pheedback https://api.pheedback.co/mcpChatGPT
Search for Pheedback in the ChatGPT connector marketplace and click Connect. You'll be redirected to Pheedback to authorize access, then returned to ChatGPT with the tools ready to use.
Authentication
The server uses OAuth 2.0. When you connect Pheedback through Claude or ChatGPT's connector marketplace, you'll be redirected to Pheedback to sign in and authorize access. Your client stores the resulting token automatically — there's nothing to copy, paste, or manage manually.
Tools
Each tool is called by the AI automatically based on context. You can also invoke them explicitly by name in your prompt.
list_organizations List all organizations the authenticated user belongs to. Always call this first — you need the organization ID for every other tool.
Parameters
None required.
Returns
An array of organization objects, each with an id (UUID) and name.
What Pheedback organizations do I have access to?list_pheedbacks List all pheedbacks (surveys/experiences) for a given organization.
Parameters
| Name | Type | Required | Description |
|---|---|---|---|
organizationId | string (UUID) | Required | Organization ID from list_organizations |
Returns
An array of pheedback objects including their IDs, names, goals, and status.
Show me all the surveys in my organization.create_pheedback Create a new pheedback (survey/experience) for an organization. Use list_organizations first to find the organization ID.
Parameters
| Name | Type | Required | Description |
|---|---|---|---|
organizationId | string (UUID) | Required | Organization ID from list_organizations |
name | string | Required | Display name for the pheedback |
goal | string | Required | The expected outcome you want to achieve from this pheedback |
Returns
The created pheedback object including its new id, embed snippet, and share URL.
Create a new survey called "Post-onboarding check-in" with the goal
of understanding where new users get confused in the first week.get_pheedback_responses Get visitor responses and answers for a specific pheedback.
Parameters
| Name | Type | Required | Description |
|---|---|---|---|
organizationId | string (UUID) | Required | Organization ID from list_organizations |
experienceId | string (UUID) | Required | Pheedback ID from list_pheedbacks |
limit | number | Optional | Max visitors to return. Default: 20, max: 20. |
offset | number | Optional | Pagination offset. Default: 0. |
Returns
A paginated list of visitor response objects, each containing the visitor's answers, timestamps, and any adaptive follow-up exchanges.
Summarize the last 20 responses from my NPS survey and identify
the top themes in the negative scores.Usage examples
Research workflow
A typical AI-assisted research session flows through the tools in order:
1. "What organizations do I have access to?"
→ list_organizations → returns org ID
2. "Show me all my active surveys."
→ list_pheedbacks(organizationId) → returns survey list
3. "Pull the last 20 responses from the churn survey and summarize
the top reasons users are leaving."
→ get_pheedback_responses(organizationId, experienceId)Creating a survey from a research question
"We're launching a new pricing tier next week. Create a Pheedback
survey called 'Pricing feedback – June launch' with the goal of
understanding which features users most associate with premium value."The AI will call list_organizations to resolve your org ID, then call create_pheedback with the name and a goal it drafts from your request.
Paginating through large response sets
The limit parameter caps at 20 per call. To page through more results, increment offset:
"Get responses 21–40 from the onboarding survey."
→ get_pheedback_responses(organizationId, experienceId, limit=20, offset=20)Troubleshooting
401 Unauthorized
Your OAuth token has expired or been revoked. Disconnect and reconnect the Pheedback integration in Claude or ChatGPT to re-authorize. If the problem persists, contact .
The AI isn't calling Pheedback tools automatically
Try being explicit: "Use the Pheedback MCP to list my surveys." If tools still don't appear, confirm the server is listed under connected MCP servers in your client and restart the app.
"Organization not found" or empty list from list_pheedbacks
Always call list_organizations first to get a valid UUID — do not hard-code or
guess organization IDs. UUIDs change between environments (staging vs. production).
Responses are truncated at 20
This is a hard server limit per call. Use the offset parameter to page through
additional results: offset=0 for the first page, offset=20 for the
second, and so on.