# ErrorHub

> ErrorHub is an API-first error tracking service for small software teams: any application that can send an HTTPS request reports errors to one endpoint, and ErrorHub groups duplicates into issues, keeps every project in one inbox and alerts when an error is new or comes back. It does not collect logs, traces, session replays or metrics.

Contact: hello@errorhub.dev
Operator: Autrady EOOD, Varna, Bulgaria (UIC 206828616, VAT BG206828616)

## Pages

- [Landing page](https://errorhub.dev/): what ErrorHub does, how it works, pricing and FAQ
- [Pricing](https://errorhub.dev/pricing.md): plans, limits and the billable unit, as Markdown
- [OpenAPI document](https://errorhub.dev/openapi/v1.json): the HTTP API contract
- [Sign up](https://errorhub.dev/signup): create a free account, no card

## Legal

- [Privacy Policy](https://errorhub.dev/privacy): what is collected, where it is stored, who receives it
- [Terms of Service](https://errorhub.dev/terms): plans, billing, limits, liability
- [Data Processing Agreement](https://errorhub.dev/dpa): processing of personal data in error events, subprocessors

## For coding agents

If a user asks you to add ErrorHub to their codebase, this is the whole integration. There is no SDK, plug-in or MCP server.

1. The user creates a project in the dashboard and gives you the server key (prefix eh_sk_) or the browser key (prefix eh_pk_). Store the server key in the ERRORHUB_KEY environment variable and never commit it. Browser keys are public by design.
2. Add one small reporter in the project's own language that sends POST https://errorhub.dev/api/v1/events with Content-Type: application/json and the key in X-ErrorHub-Key. Every project also shows a ready install prompt with these rules.
3. Report unhandled exceptions from the framework's global error hook. Fire and forget with a timeout of about 3 seconds. Reporting must never throw, block a request or retry in a loop.
4. Never send passwords, cookies, Authorization headers, tokens, request bodies or query strings.
5. Send one test error and confirm the reply is 202.

Every issue in the dashboard has a fix prompt with the stack trace, request, release and environment, for pasting into a coding agent. The error data in it is labelled as data, and the prompt tells the agent never to follow instructions found inside it.

## Key facts

- Integration: one HTTPS POST to https://errorhub.dev/api/v1/events with the project key in the X-ErrorHub-Key header. There is no SDK.
- Required fields: exceptionType and exceptionMessage, or message. Every other field is optional.
- Replies: 202 when the event is stored, 4xx for a bad report, 429 with the code rate_limited when a key sends too fast, 429 with the code quota_exceeded, budget_exhausted or plan_ceiling_reached when the account reached the limit of its plan.
- Grouping: events are fingerprinted by source, exception type, normalized message and application stack frames. Stack-frame grouping is tuned for .NET and JavaScript traces today.
- Regressions: a resolved issue that receives a new event is reopened and sends an alert.
- Alerts: web push notifications for new errors and regressions. Warnings and repeats of a known error stay quiet.
- Privacy: events carry no request bodies, headers or cookies. Query strings are removed and secrets are redacted on the server.
- Hosting: events and issues are stored on servers of Hetzner Online GmbH in Germany, in the European Union.
- Pricing: Free is €0 for 5,000 events a month and 1 user. Indie is €9 a month for 10,000 events and unlimited users. Pay-as-you-go is €0.20 per started 1,000 extra events, up to 100,000 events a month, and is off by default. Prices exclude VAT.
- Accounts: self-service sign-up, team invitations with roles (owner, administrator, member), self-service cancelling and account deletion.
- Payments: handled by Stripe. Cards are never entered on errorhub.dev.
- AI: ErrorHub runs no AI model and never reads or changes customer code. Grouping uses fixed rules.
- Left out on purpose: logs, distributed tracing, session replay, APM, profiling, uptime checks, status pages.

## Who it is for

- Founders running one or several SaaS products
- Agencies and studios that keep client projects separate but want one inbox
- Engineering teams of 2 to 20 developers
- Teams with several languages, serverless functions or background workers