Catch every error
before your users do

Track errors in your apps with a lightweight SDK. European-hosted, privacy-first, and priced for indie hackers.

Get started for free
Free 14-day trial·Set up in 2 minutes·GDPR compliant
ObserverKit dashboard

Simple pricing, no hidden fees

Start for free. Upgrade only when your project grows.
Two plans. No confusion.

0€/ month

No credit card needed.

  • 500 errors / month
  • Unlimited projects
  • 30-day retention
  • Telegram notifications
  • Slack notifications
  • Email notifications
  • Email support
  • AI fix proposals
10€/ month

14-day free trial included.

  • 25,000 errors / month
  • Unlimited projects
  • 90-day retention
  • Telegram notifications
  • Slack notifications
  • Email notifications
  • Email support
  • AI fix proposals, 20 per month

We're independent and do not answer to investors.

Frequently asked questions

How do I connect ObserverKit to Claude Code?

Create an API key in your project settings, then run: claude mcp add --transport http observerkit https://api.observerkit.com/mcp --header "Authorization: Bearer YOUR_API_KEY". Claude Code can then read and triage your errors. Any MCP client that speaks streamable HTTP works the same way. Full walkthrough in the agents docs.

Can an AI agent fix my errors automatically?

Yes, with your own coding agent. ObserverKit exposes your errors over MCP, so an agent like Claude Code can pull the day's new errors, fix them on a branch, run your tests, and open the pull request in your repository. The fixing and the pull request happen in your agent and your repo, not in ObserverKit. If you would rather not wire that up, the dashboard's AI fix proposal reads your linked GitHub repo and hands you an explanation and a diff to copy.

What can the MCP server actually do?

It exposes five tools, all scoped to the project that owns your API key: list_error_groups to list errors with status, time and service filters; get_error_group for a symbolicated stack trace with source context; list_error_events for individual occurrences; resolve_error_group to close an error; and ignore_error_group to mute noise.

Do I need a paid plan for the MCP server or the API?

No. The MCP server and the REST API are available on every plan, including Free. Only AI fix proposals in the dashboard require the Pro plan.

How is ObserverKit different from Sentry?

Three things: it is priced for indie developers and small teams rather than enterprises, it is hosted in Europe and GDPR compliant by default, and it is built for agents. The MCP server and REST API are first-class, so your coding agent can read, triage and resolve errors without a human in the loop.

Where is my data stored?

In Europe. ObserverKit is European-hosted and GDPR compliant, and error events are subject to your plan's retention window.

What frameworks does the SDK support?

@observerkit/js is framework-agnostic: it works in any JavaScript or TypeScript project by calling ObserverKit.init. The docs ship step-by-step guides for React, React Router and Next.js, and there is a Vite plugin for uploading source maps.