Skip to main content

Gather user feedback that fixes itself.

Your users report bugs. Nobody responds. You're too busy shipping to play support hero. One line adds a feedback widget and MCP Server — your AI coding agent aggregates feedback, opens PRs, and drafts responses. You stay in control.

Add to your app in one command

$ npx userdispatch init
See exactly what this command does →

Under 30KB · 2 min setup · Shadow DOM isolated · No lock-in

Send Feedbackpowered by UserDispatch
Drop files anywhere or click to attach

The smart feedback loop

Turn user feedback into pull requests

UserDispatch closes the loop between your users and your app. No dashboards to check. No tickets to triage. Your agent analyzes feedback and proposes changes — you review and merge.

01

Your user submits feedback

Bugs, ratings, questions, screenshots, and app logs captured automatically via the widget.

02

Your coding agent reads it instantly

Your coding agent pulls new submissions via the UserDispatch MCP Server. No polling, no webhooks, no manual checking.

03

Your agent analyzes and acts

Aggregates patterns across submissions, opens a PR with the fix, and drafts a response for your review.

04

Weekly digest

Built-in prompt generates trend reports, top issues, and resolution rates — without opening a dashboard.

Why now

Support tools were built for a pre-agent world

Canny, Intercom, and Zendesk assume a human reads every ticket and weren't built AI/MCP-first. In the agentic era, your AI coding agent should be the first responder.

Legacy tools

  • Dashboard for humans to read and sort tickets
  • Manual triage, tagging, and prioritization
  • Human writes the fix, human responds to user
  • API bolted on as afterthought, no agent protocol
  • $50–300/mo for teams, complex setup

UserDispatch

  • MCP server for agents to read submissions natively
  • Agent triages, categorizes, and spots patterns across submissions
  • Agent opens a PR with the fix and drafts a reply for your approval
  • MCP-first: tools, resources, and built-in prompts for your agent
  • Free to start. $0 to prove the concept. You keep building — your agent handles the analysis.

Ready to try it?

$ npx userdispatch init

Free · 2 minutes · No credit card

What you get

Everything to close the loop

A lightweight widget for your users. A powerful MCP server for your agent. Nothing else to manage.

3 feedback types

Ratings, bug reports, and questions — all in a single widget your users already understand.

Screenshots & uploads

File attachments and screenshots with full browser metadata: OS, viewport, URL, and user agent.

Application logs

Pipe console errors, exceptions, and custom events via the API. Your agent sees what your user saw.

Full MCP toolkit

Lifecycle management for submissions, apps, stats, org, and members. Correlate with Sentry errors, failed GitHub deploys, and other data sources.

Agent-sent emails

Your agent drafts responses to users.

Custom UI

Personalize your styles, dark mode support, custom accent colors and logo. Looks native in any app.

MCP-first

Built for AI agents, not dashboards

Connect the UserDispatch MCP server to Claude Code, Cursor, or any MCP-compatible agent.

Triage incoming

Review new submissions, categorize by severity, flag critical bugs instantly.

Reply to users

Draft email responses for your review. Your users get answers in minutes, not days.

Track resolution

Update statuses, close resolved items, and track the full feedback lifecycle.

Weekly digest

Built-in prompt generates trends, top issues, and resolution rates automatically.

.mcp.jsonMCP
// Add this to your .mcp.json — that's it
{
  "mcpServers": {
    "userdispatch": {
      "url": "https://userdispatch.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ud_..." // or sk_ for app-scoped access
      }
    }
  }
}

tools · resources · promptssubmissions, apps, stats, org, members

submissionsappsstatsorgmemberstriage-submissionsweekly-digest

Comparison

Not another support ticketing system

Legacy tools were built for humans reading every ticket. UserDispatch was built for agents analyzing them and proposing fixes.

FeatureLegacy support toolsUserDispatch
One-prompt setup
Native MCP server for AI agents
Built-in log capturing~
Agent-driven feedback-to-PR pipeline
Manage multiple vibe-coded apps in one place

Security

Enterprise-grade security, built in

Every layer — from widget to database — is hardened. Five rounds of security audits, 110+ findings remediated. No enterprise plan required.

Hashed API tokens

SHA-256 at rest, 90-day expiry, shown once at creation. Raw tokens are never stored.

Nonce-based CSP

Per-request Content-Security-Policy with strict-dynamic. No unsafe-inline in production.

Multi-tier rate limiting

PostgreSQL-backed atomic counters with per-IP, per-key, and global limits across all endpoints.

File content validation

Magic-byte verification blocks disguised HTML, SVG, and XML uploads regardless of extension.

Parameterized queries

Tagged templates with bound parameters on every database query. Zero string interpolation in SQL.

Role-based MCP access

Auth context on all tools with org-scoped data isolation. Owner-only destructive operations.

Get started

Two minutes. One command.

Your coding agent handles the rest.

01
npx userdispatch init

Install

Authenticates you, creates your app, injects the widget script tag, and configures your MCP server.

Full breakdown →
02
.mcp.json

Connect your agent

Claude Code, Cursor, and any other MCP-compatible agent can act on user feedback instantly.

03
Review & merge

Review what your agent proposes

Your agent triages feedback, opens PRs with fixes, and drafts responses — you review and ship.

Already have an account? Sign in

Built on the Model Context Protocol

Designed for the tools you already use

UserDispatch works with any MCP-compatible coding agent. Set up once, use everywhere.

Claude Code
Cursor
OpenAI Codex
Windsurf
Any MCP Agent

MCP tools · resources · built-in prompts · Bearer token auth

Pricing

Simple pricing. Start free.

Free

$0forever

Ship your first app with feedback that works on day one.

  • 100 submissions/month
  • 1 user
  • 2 apps
  • 3 month data retention
  • Hosted MCP server
  • Feedback widget
  • Agent-sent emails
  • App log capture
Get started
Recommended

Pro

$9/mo per workspace

Your app has users. Now keep them.

  • 250 submissions/month
  • 1 user
  • 10 apps
  • Unlimited data retention
  • Everything in Free
  • Custom branding & logo
  • Priority support
  • File uploads up to 25MB
Get started

Team

$49/mo per workspace

Your team moves quickly. Your feedback should too.

  • 2,500 submissions/month
  • 5 users
  • Unlimited apps
  • Unlimited data retention
  • Everything in Pro
  • Team roles & permissions
  • Advanced analytics
Get started

Enterprise

Custom

For organizations building at scale with full autonomy.

  • Unlimited submissions
  • Unlimited users
  • Unlimited apps
  • Everything in Team
  • Autonomous product roadmap & priorities
  • SSO & advanced security
  • Custom forms
  • Dedicated account manager

Gather user feedback that fixes itself.

One command. Two minutes. Your coding agent analyzes feedback and opens PRs — so you can keep building.

$ npx userdispatch init
See what this does →

FAQ

Common questions

What is UserDispatch?
UserDispatch is a feedback widget and MCP server that lets your AI coding agent analyze user feedback and propose fixes. Your users submit bugs, ratings, and questions through the widget. Your agent aggregates patterns, opens PRs, and drafts responses — you review and merge.
What frameworks does the widget support?
The widget works with any web framework: Next.js (App Router and Pages Router), Vite, Create React App, Nuxt, SvelteKit, Astro, and plain HTML. The CLI auto-detects your framework and injects the script tag in the right place.
Do I need to write any code?
No. Run npx userdispatch init and the CLI handles authentication, widget injection, and MCP configuration. The entire setup is interactive and takes about two minutes.
What coding agents are supported?
Any MCP-compatible agent works: Claude Code, Cursor, Windsurf, OpenAI Codex, and more. The CLI auto-detects your agent and writes the MCP config file. You can also connect manually via the cloud MCP endpoint.
How is this different from Canny or Intercom?
Those tools were built for humans reading tickets. UserDispatch was built for AI agents analyzing them and proposing fixes via PRs. Your coding agent gets a native MCP server — not a REST API bolted on as an afterthought.
Is my users’ data safe?
Yes. API tokens are SHA-256 hashed at rest, all queries use parameterized SQL, file uploads are validated by magic bytes, and every MCP tool enforces org-scoped access control. We run nonce-based CSP, multi-tier rate limiting, and constant-time token comparison. Read the full details
What’s included in every tier?
Every tier includes the full hosted MCP server, all tools, and the feedback widget. We never drop submissions — hit your limit and we still capture everything. Upgrade anytime to access them.