Gamebeast for Web
Agents that run your web analytics
Gamebeast's agent sets up your tracking in the browser and on your server, finds what's moving and why, tests a fix in an experiment, and ships what wins. Every session is tied to the campaign it came from.
What the agent runs on Web
Browser and server SDKs
TypeScript SDKs for browsers, Node.js, Bun, Deno, and edge runtimes.
Attribution per session
Every session tied to the deeplink or UTM campaign it came from.
Server-side configs
Evaluate configs for any user from your backend, and render safely during SSR.
Experiments (A/B testing)
Tune values in a live build with remote configs, and prove changes before they ship.
Your agent sets it up
Run /instrument and the agent writes your Web tracking with your coding agent, then checks each event as it arrives. Prefer to install it yourself? Here’s how.
- 1
Install @gamebeast/js for the browser, and @gamebeast/node for your backend if you need it.
- 2
Create an SDK key for browser code, and a server key for your backend, in Project Settings in the Gamebeast dashboard.
- 3
Create one client per page and send your first event.
- Full Web installation guide
import { GamebeastClient } from "@gamebeast/js";
export const gamebeast = new GamebeastClient({
apiKey: "YOUR_SDK_KEY",
});
gamebeast.markers.send("page_opened", { path: location.pathname });The agent does the work, not you
Most analytics tools stop at a chart. Gamebeast's agent does the job of an analyst and a LiveOps lead: it instruments your app, website, or game, finds what's moving and why, and changes your remote configs and experiments to act on it. Use it in our dashboard, or connect Claude, Cursor, or Codex over MCP at no extra cost.
- Step 1
Instrument. Type
/instrument. The agent plans your events and the values worth testing, hands the code to your coding agent, and checks every event as it arrives. - Step 2
Find. It watches your data like an analyst: why a metric moved, where users drop or spend, which experiment is ready for a decision, what your community is complaining about.
- Step 3
Act. It builds the dashboard and the cohort, tunes your remote configs, then designs, launches, ramps, and calls the experiment. You approve every change before it ships.
Plus a playbook for the rest of the job
/diagnoseWhy a metric moved/weekly-reviewWeekly review/decide-experimentsShip or kill tests/economyTune the economy/liveopsLaunch a LiveOps event/segmentTarget a segment/player-voiceAct on user feedback/dashboardsBuild team dashboards/fix-instrumentationFix your data
SDKs: Roblox · Unity · Unreal · Godot · JavaScript · Node · HTTP API
