MCP Server Documentation
The official @feedjar/mcp package provides a Model Context Protocol (MCP) server for FeedJar. It allows AI coding assistants like Cursor, Claude Desktop, Antigravity, and Windsurf to list and submit in-app feedback for your project directly inside your IDE environment.
Installation
You can run the MCP server on demand via npx or install it globally:
npx -y @feedjar/mcpAuthentication & Environment Variables
The MCP server authenticates using an app ingest API key (fjr_…) generated from your FeedJar Dashboard under App Settings → Ingest Key.
| Variable | Required | Default | Description |
|---|---|---|---|
FEEDJAR_API_KEY | Yes | — | App ingest key from the FeedJar dashboard (fjr_…) |
FEEDJAR_API_BASE_URL | No | https://api.feedjar.in | API base URL (no trailing slash) |
Available MCP Tools
1. list_feedback
Retrieves incoming user feedback, feature requests, support tickets, and crash reports for the keyed app (newest first).
// Tool Arguments
{
"limit": 20, // Optional: max items (default 50, max 200)
"type": "feature" // Optional: "feedback" | "feature" | "support" | "crash"
}2. submit_feedback
Submits a new feedback item or test report for the app.
// Tool Arguments
{
"message": "User reported crash on swipe back",
"type": "crash",
"email": "dev@example.com", // Required if type is "support"
"rating": 5, // Optional: 1–5 score
"metadata": { "screen": "cart", "appVersion": "1.4.0" }
}Client Configurations
Cursor Setup
Add to your project's .cursor/mcp.json file or open Cursor Settings → MCP → Add New MCP Server:
{
"mcpServers": {
"feedjar": {
"command": "npx",
"args": ["-y", "@feedjar/mcp"],
"env": {
"FEEDJAR_API_KEY": "fjr_your_app_ingest_key"
}
}
}
}Claude Desktop Setup
Add to claude_desktop_config.json:
{
"mcpServers": {
"feedjar": {
"command": "npx",
"args": ["-y", "@feedjar/mcp"],
"env": {
"FEEDJAR_API_KEY": "fjr_your_app_ingest_key"
}
}
}
}Next Steps
- Learn more about FeedJar MCP Features.
- Read the Feedback API Reference.
- Check the Getting Started Guide.