Self-hosted service
rybbit-io/rybbit avatar
rybbit-io/rybbit

Rybbit: self-hosting a ClickHouse, Postgres and Redis web analytics stack under AGPL-3.0

Rybbit - open-source and privacy-friendly alternative to Google Analytics that is 10x more intuitive.

13,319 stars784 forksTypeScriptAGPL-3.0

At a glance

What is it?
Rybbit is an open-source, cookieless web and product analytics platform with session replay, funnels and maps. Hosting it yourself means running several services whose sample credentials are weak, and some features, such as Web Vitals, are limited to the paid cloud.
Who is it for?
Rybbit fits teams that want cookieless analytics with session replay and funnels and are ready to operate ClickHouse, Postgres and Redis themselves, or to pay for the hosted service. It is a weaker choice if you need Web Vitals without paying, or if you would deploy the sample environment file unchanged, because its default passwords and open signup are unsafe on a public host.
Can I use it commercially?
Yes, with strict conditions. AGPL-3.0 is a network copyleft licence: if people use a modified version over a network, for example as a hosted service, you must offer them its source code under the same licence.
Is it still maintained?
Yes. The repository last received commits 2 days ago.
What is it written in?
Mainly TypeScript, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • rybbit-io/rybbit13,319 ★ · +786
Star history
Repository2026-07-282026-11-15
rybbit-io/rybbit12.5k13.3k

786 Stars in the last 110 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

Self-hosting means running ClickHouse, Postgres, Redis and two app services

Rybbit offers two routes: a hosted service, described as the fastest way to start, and self-hosting on your own VPS for complete control. The compose file shows what the second route involves. It defines a ClickHouse server, a backend, a client, and an optional Caddy web server enabled through a profile named with-webserver. The environment file also configures Postgres and Redis.

So the install is a multi-service deployment, not a script tag and a binary. ClickHouse stores the analytics data and has its own configuration files for network, logging and resource limits, mounted as configs. Plan memory and disk for ClickHouse in particular, and decide who will watch the databases, because the project's headline claim of setting up in a couple of minutes describes the best case.

The sample credentials are frog and changeme, and signup is open by default

The example environment file sets the ClickHouse password to frog, the Postgres user and password both to frog, and the Redis password to changeme. It requires BETTER_AUTH_SECRET and tells you to generate it with openssl rand -hex 32. It also sets DISABLE_SIGNUP to false.

code
BETTER_AUTH_SECRET=replace-with-output-of-openssl-rand-hex-32
HOST_BACKEND_PORT=127.0.0.1:3001:3001
HOST_CLIENT_PORT=127.0.0.1:3002:3002

The consequence is concrete. Copying the file unchanged onto a public server leaves the databases with guessable passwords, and with signup enabled anyone who finds the dashboard can create an account. Replace every password, generate the secret, and set DISABLE_SIGNUP to true after you create your own user if you do not want strangers registering.

Backend and client ports bind to localhost, but Caddy publishes 80 and 443

The host port variables default to loopback: the backend on 127.0.0.1 port 3001 and the client on 127.0.0.1 port 3002. The file explains the format, host IP, host port, container port, and warns that without a host IP the port is published on all interfaces. Removing the address prefix therefore exposes the service to the network.

The Caddy service is separate. When enabled, it publishes ports 80 and 443, plus 443 over UDP for HTTP/3, and reads a Caddyfile with the domain from DOMAIN_NAME. The USE_WEBSERVER setting turns the built-in web server off, which is the option to take if you already run your own reverse proxy and want to keep it in front of the loopback ports.

ClickHouse access management and the fixed default user

The compose file pins ClickHouse to a specific server image and sets CLICKHOUSE_DEFAULT_ACCESS_MANAGEMENT to 1. A comment says this lets the backend provision a least-privilege rybbit_query user, used for custom SQL queries and dashboards, with SQL at startup.

There is a quirk to respect in the environment file: the backend always connects to ClickHouse as the default user, so you should not set CLICKHOUSE_USER. If you change the user name and expect the backend to follow, it will not. The password is the only ClickHouse credential you control, and since access management is enabled for the default user, that password protects more than data reads. Treat it as an administrative secret.

The comparison table is the project's own, and Web Vitals is cloud only

The README compares Rybbit with GA4, Plausible and Cloudflare across about twenty features, and Rybbit comes out ahead in most rows, including advanced maps, session replays, user profiles and error tracking. This is a self-assessment, so weigh it accordingly. Two footnotes carry real limits: Plausible's Community Edition has very limited features compared to its cloud version, and Web Vitals for Rybbit is only available on paid tiers.

The dashboard preview section repeats the point with a heading marking Web Vitals as cloud only. If you self-host to avoid paying, expect to go without Web Vitals. The remaining features listed are sessions, unique users, pageviews, bounce rate, session duration, goals, retention, user journeys, funnels, filtering across more than 15 dimensions, custom events with JSON properties, and country, region and city location tracking.

AGPL-3.0 licensing applies to anyone who modifies and hosts it

Rybbit is licensed under AGPL-3.0, and the README links to the license from the top navigation. AGPL is a strong copyleft license, so if you modify the code and run it for users over a network, you should read the license terms about offering the corresponding source before you deploy changes.

The license also affects commercial planning. A company that wants to fold Rybbit into a closed product cannot treat it like an MIT-licensed component. Using it unmodified for your own sites is a different situation from redistributing a changed version. The hosted service exists for people who would rather not manage any of this, and the repository includes SECURITY.md for reporting vulnerabilities.

A pnpm monorepo that needs Node 22.13 or newer

For local development the repository is a pnpm workspace, with the package manager pinned to pnpm 10.33.0 and an engines field requiring Node 22.13.0 or newer. The root scripts build a shared package first, then run client and backend in parallel for dev, with separate dev:client and dev:server entries and matching build scripts. The test script runs workspaces one at a time with the workspace concurrency set to 1, so a full run is slower than the parallel dev scripts.

The tree also holds react-native, clickhouse, docs and ops directories, plus design and context files for contributors. The recent releases are v2.9.2 on 2026-10-24, v2.9.0 on 2026-09-12 and v2.8.0 on 2026-07-27, and the last push was on 2026-11-02.

Editorial conclusion

Rybbit fits teams that want cookieless analytics with session replay and funnels and are ready to operate ClickHouse, Postgres and Redis themselves, or to pay for the hosted service. It is a weaker choice if you need Web Vitals without paying, or if you would deploy the sample environment file unchanged, because its default passwords and open signup are unsafe on a public host. Check the AGPL-3.0 terms and change every credential before the first deploy.

Frequently asked questions

what is rybbit

Rybbit is an open-source, privacy-friendly web and product analytics platform with session replay, funnels, goals and maps, available as a hosted service or for self-hosting.

rybbit alternative

The project positions Rybbit as an alternative to Google Analytics. Its README compares it with GA4, Plausible and Cloudflare Web Analytics on features such as session replay and error tracking.

Is Plausible free to use?

The README marks Plausible as self-hostable, with a footnote that its Community Edition has very limited features compared with the cloud version. It gives no pricing.

What is the best free analytics tool?

The README does not rank tools. Rybbit is free to self-host under AGPL-3.0, but you run the servers yourself, and Web Vitals is only available on paid tiers.

Official sources

  1. Official documentation
  2. Official README
  3. Project repository
  4. Release notes
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/rybbit-io-rybbit.svg)](https://hysenlabs.com/projects/rybbit-io-rybbit)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/rybbit-io-rybbit.svg)](https://hysenlabs.com/projects/rybbit-io-rybbit)

Was this useful?