Remote UserDispatch MCP Server

The MCP server gives AI assistants full access to your submissions, apps, organization settings, and statistics. It's the most powerful way to manage UserDispatch — your agent handles the entire feedback lifecycle.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Think of it as a USB port for AI — it provides a standard way for Claude (and other assistants) to interact with services like UserDispatch. Once connected, you can ask Claude to triage your inbox, draft replies, pull stats, and more, all in natural language.

Setup by agent

The easiest way is npx userdispatch init — it auto-detects your agent and writes the config. Or configure manually:

Create .mcp.json in your project root:

.mcp.json

{
  "mcpServers": {
    "userdispatch": {
      "url": "https://userdispatch.com/api/mcp",
      "headers": {
        "Authorization": "Bearer your-api-token"
      }
    }
  }
}

Restart Claude Code after saving.

Authentication

The MCP server uses Bearer token authentication. There are three ways to set it up, from easiest to most advanced:

Automatic setup (recommended)

Run npx userdispatch init — it generates a ud_ token, writes it to your MCP config file, and configures everything automatically. This is the easiest approach and what most users should do.

Manual MCP client config

If you prefer to configure manually, you only need a ud_ token in the Authorization header. To get a token:

Go to userdispatch.com/cli/auth
Sign in with Google
Copy the ud_ token

Paste it as the Bearer token in your MCP config (see the agent-specific examples above). Your org is resolved server-side from the token — you don't need to set an org ID.

Tools reference

Each tool uses an action parameter to select the operation. Action-specific fields are validated at runtime.

Tool	Description	Actions	Parameters
submissions	Manage user submissions — list, view details, update status, reply, or delete	list, get, update, reply, delete	action, id?, status?, body?, channel?, app?, type?, since?, search?, limit?, offset?
apps	Manage apps/projects — list, create, update settings, delete, or rotate API key	list, create, update, delete, rotate_key	action, id?, name?, slug?, notification_email?
stats	Get submission statistics, optionally filtered by app and time period	—	app?, days?
org	View or update organization settings	get, update	action, name?, slug?, notification_email?
members	Manage organization members — list, invite, update role/settings, or remove	list, invite, update, remove	action, id?, email?, name?, role?, notify_email?

Resources reference

URI	Description
userdispatch://submissions/recent	Last 20 submissions across all apps
userdispatch://submissions/{id}	Full details of a specific submission
userdispatch://stats	Current submission statistics overview
userdispatch://apps	All registered apps with configuration
userdispatch://org	Organization details and member list

Prompts reference

triage-submissions

Reviews all new/unresolved submissions and suggests actions for each: priority level, status change, draft replies, and spam/duplicate detection. Great for a daily inbox review.

weekly-digest

Generates a 7-day summary with key metrics, trends, breakdowns by type and status, notable items, and recurring patterns. Perfect for weekly team updates.

Example workflows

Once connected, ask your AI agent things like:

Triage your inbox

“Review my new submissions and prioritize them. Mark any spam as closed and draft replies for the high-priority bugs.”

Weekly digest

“Generate a weekly digest of all feedback from the last 7 days. Include trends and any recurring issues.”

Reply to users

“Find all unresolved bug reports from the last week and send a reply to each one letting them know we're investigating.”

Manage apps

“Create a new app called ‘Mobile App’ with slug ‘mobile-app’, then show me the API key.”

Troubleshooting

MCP tools not appearing

Make sure you restarted your AI agent after saving the config file. Check that the config path is correct for your agent.

Authentication failures

Verify your ud_ token is correct and the header format is Authorization: Bearer your-token. Tokens generated by the CLI are already valid.

Server not responding

The MCP endpoint is https://userdispatch.com/api/mcp. Make sure you're using HTTPS, not HTTP.