Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Feedbucket MCP Servershow me the 5 most recent unresolved feedback items"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Feedbucket MCP Server
A Model Context Protocol (MCP) server that lets AI assistants (Claude Code, Cursor) manage website feedback from Feedbucket.
Beta Software: This project is in beta. Use at your own risk. Please report issues on GitHub.
Prerequisites
Before installing this MCP server, make sure you have:
A Feedbucket account
Feedbucket installed on your website
Protected mode enabled (recommended):
Go to Widget Settings
Set "Trigger Feedbucket using a query string" to Yes
Save settings and copy your secret key
Quick Start
1. Install from npm
2. Run the setup wizard
The setup script automatically:
Extracts the project ID from your website HTML
Fetches the private key from Feedbucket API
Configures Claude Code and Cursor
3. Restart your IDE
Restart Claude Code or Cursor to load the new MCP server.
4. Start using it
Ask your AI assistant:
"Show me unresolved feedback from Feedbucket"
"What's the feedback summary for this project?"
"Add a comment to feedback #12345 saying we've fixed this"
Alternative: Clone from GitHub
If you prefer to clone the repository:
Available Tools
Tool | Description |
| List and filter feedback with smart summarization |
| Get full details of a specific feedback item |
| Quick project overview (total, resolved, by type, recent activity) |
| Add comments to feedback items |
| Mark feedback as resolved |
| Check API connection status |
feedback_list
List feedback with powerful filtering:
feedback_get
Get complete details for a single feedback item:
Returns full text, all comments, attachments, browser/device info, and console logs.
feedback_stats
Get a quick project health overview:
feedback_comment
Add a comment to a feedback item:
feedback_resolve
Mark a feedback item as resolved:
Setup Options
Automatic Setup (recommended)
Manual Project ID (if auto-detection fails)
Configure only one IDE
Extract credentials without configuring
Full help
Manual Configuration
If you prefer to configure manually:
Claude Code
Add to .mcp.json in your project root:
Or use the CLI:
Cursor
Add to .cursor/mcp.json in your project root:
Environment Variables
Variable | Required | Description |
| Yes | Project ID from the |
| Yes | Private key for commenting/resolving (auto-fetched by setup) |
| No* | API key for protected projects ( |
*Required if your project uses query string protection.
Development
Architecture
This MCP server was reverse-engineered from the Feedbucket browser widget to provide:
Smart Summarization: Optimizes large datasets for AI consumption
Client-side Filtering: Reduces API calls by filtering locally
Full TypeScript: Strict mode with comprehensive types
Production Ready: Proper error handling and validation
API Endpoints Used
Method | Endpoint | Purpose |
GET |
| Fetch project with all feedback |
POST |
| Add comment to feedback |
PUT |
| Mark feedback as resolved |
License
MIT