walkerOS Flow MCP Server
io.walkeros/flow
Build, validate, simulate, and deploy event pipelines with config-as-code tracking that you own and version.
What is the walkerOS Flow MCP server?
The walkerOS Flow MCP server enables AI agents to work with walkerOS event pipelines—a warehouse-native event collection layer that replaces scattered GA4, custom tags, and ad pixels with a single, versioned configuration. It allows reading event schemas, validating pipeline definitions, simulating events, and generating integration code, making your event tracking programmable via AI.
walkerOS Flow is an MCP interface to walkerOS, an open-source event collection platform that centralizes tracking from browser to data warehouse and ad platforms. Instead of managing separate GA4, custom tags, and ad pixels, you define events once in config-as-code, version them in git, validate them at collection time, and route them to multiple destinations. The MCP server lets Claude and other AI agents read your schema, suggest tracking definitions, validate configurations, and generate code—turning event pipelines into something AI can help you build and maintain.
How to install walkerOS Flow
Copy-paste configuration for popular MCP clients.
WALKEROS_TOKENsecretwalkerOS API token. Optional, the local tools work without it and the auth tool can log in instead.
WALKEROS_PROJECT_IDDefault walkerOS project id (proj_...) used by the cloud tools.
WALKEROS_APP_URLBase URL override for the walkerOS app. Defaults to https://app.walkeros.io.
Tools & capabilities
Tools this server exposes to the agent.
Load flow configuration— Read and parse walkerOS event pipeline configurationsValidate event schema— Validate event definitions against walkerOS schema rulesSimulate events— Test events through the pipeline locally without external callsGenerate integration code— Suggest tracking definitions and generate code for sources and destinations
Use cases
- Design and validate event schemas before deploying to production
- Simulate events through your pipeline to catch schema errors early
- Generate boilerplate code for connecting new data sources or destinations
- Review and version-control your entire tracking setup in git PRs
- Migrate from Google Tag Manager or scattered pixel implementations to a unified event layer
walkerOS Flow MCP server FAQ
walkerOS Flow is an MCP server that gives AI agents access to walkerOS event pipelines. It lets you build, validate, simulate, and deploy event collection configurations that route data from your app to warehouses and ad platforms.
Yes, walkerOS is MIT licensed and open-source. You can self-host it anywhere with no vendor lock-in.
In Claude Code, use `/plugin marketplace add elbwalker/walkerOS` and `/plugin install walkeros@elbwalker`. For other MCP clients, add the server to your config with `npx @walkeros/mcp`.
No. Loading flows, validating them, and simulating events all run locally. No external account or authentication is required.
walkerOS supports GA4, Google Ads, Meta CAPI, BigQuery, and more. See the full list in the destinations documentation.
Yes. walkerOS supports both integrated mode (import directly into your TypeScript app) and bundled mode (build a standalone script from JSON config).
README (reference)
Source of truth, from the repository.
Event collection you own, version, and trust
<div align="left"> <a href="https://github.com/elbwalker/walkerOS/blob/main/LICENSE"> <img src="https://img.shields.io/github/license/elbwalker/walkerOS" alt="License" /> </a> <a href="https://www.walkeros.io/docs/"> <img src="https://img.shields.io/badge/docs-www.walkeros.io/docs-yellow" alt="walkerOS Documentation" /> </a> <a href="https://github.com/elbwalker/walkerOS/tree/main/apps/demos/react"> <img src="https://img.shields.io/badge/React_demo-blue" alt="React demo" /> </a> <a href="https://storybook.walkeros.io/"> <img src="https://img.shields.io/badge/Storybook_demo-pink" alt="Storybook demo" /> </a> <a href="https://www.npmjs.com/package/@walkeros/collector"> <img src="https://img.shields.io/npm/v/@walkeros/collector" alt="npm version" /> </a> </div>From browser to BigQuery and ad platforms - warehouse-native, and ready for AI agents via MCP.
The problem
GA4, custom tags, ad pixels - each one a separate setup, none of them agreeing on the same numbers. When something breaks, you can't see where.
walkerOS is one collection layer for all of them.
Why walkerOS?
- Config-as-code - version control your tracking, review it in PRs, deploy with confidence
- Declarative tagging - tag your UI in HTML, not scattered JavaScript
- Consent-native - events queue until consent is given, then flush correctly to every destination
- Schema validation - catch bad events at collection time, not weeks later in a dashboard
- One layer, many destinations - send to your warehouse and ad platforms from a single event definition
- Warehouse-native - events land clean and structured in your data warehouse, ready to query
- MIT licensed - self-host anywhere, no vendor lock-in
How it works

- Sources: Where events come from (browser, dataLayer, Express, AWS Lambda, GCP Functions, and more)
- Collector: The processing engine (consent, validation, mapping, routing, enrichment)
- Destinations: Where events go (GA4, Google Ads, Meta CAPI, BigQuery, and more)
Two ways to install walkerOS
Choose one based on your workflow and integration possibilities:
| Mode | Description | Best For |
|---|---|---|
| Integrated | Import directly into your TypeScript application | React/Next.js apps, TypeScript projects |
| Bundled | Build a standalone script from JSON config with npx walkeros | Static sites, Docker deployments, CI/CD |
Integrated (import into your app):
import { startFlow } from '@walkeros/collector';
const { elb } = await startFlow({
destinations: {
console: {
code: {
type: 'console',
config: {},
push: (event) => console.log('Event:', event.name),
},
},
},
});
await elb('page view', { title: 'Home' });
// -> logs: Event: page view
Wire a real source and destination once you are ready:
import { startFlow } from '@walkeros/collector';
import { sourceBrowser } from '@walkeros/web-source-browser';
import { destinationGtag } from '@walkeros/web-destination-gtag';
await startFlow({
sources: {
browser: {
code: sourceBrowser,
config: { settings: { pageview: true } },
},
},
destinations: {
ga4: {
code: destinationGtag,
config: {
settings: { ga4: { measurementId: 'G-XXX' } },
},
},
},
});
Bundled (build from JSON config):
{
"version": 4,
"flows": {
"default": {
"config": {
"platform": "web",
"bundle": {
"packages": {
"@walkeros/collector": {},
"@walkeros/web-source-browser": {},
"@walkeros/web-destination-gtag": {}
}
}
},
"sources": {
"browser": {
"package": "@walkeros/web-source-browser",
"config": { "settings": { "pageview": true } }
}
},
"destinations": {
"ga4": {
"package": "@walkeros/web-destination-gtag",
"config": {
"settings": { "ga4": { "measurementId": "G-XXX" } }
}
}
}
}
}
}
Then: npx walkeros bundle flow.json
- Operating Modes
- Quickstart guide for React
- Full Documentation - Complete guides and API reference
- Destinations - GA4, Meta, BigQuery, and more
- React Demo
- Storybook
AI-ready via MCP
walkerOS exposes a Model Context Protocol (MCP) interface. AI agents can read your event schema, suggest tracking definitions, and generate integration code - making your event layer programmable, not just configurable.
In Claude Code, one plugin installs both MCP servers and the walkerOS skills:
/plugin marketplace add elbwalker/walkerOS
/plugin install walkeros@elbwalker
For any other MCP client, add the servers to its configuration:
{
"mcpServers": {
"walkeros-flow": {
"command": "npx",
"args": ["@walkeros/mcp"]
},
"walkeros-source-browser": {
"command": "npx",
"args": ["@walkeros/mcp-source-browser"]
}
}
}
Loading a flow, validating it, and simulating an event all run locally, no account needed. See the MCP docs.
Coming from Google Tag Manager? See walkerOS vs. GTM.
Contributing
⭐️ Help us grow and star us. See our Contributing Guidelines to get involved.
Support
Need help? Start a discussion, or reach out via email.
For more insights, visit the talks repository.
License
Licensed under the MIT License.
Related MCP servers

walkerOS Source Browser
Generate, parse, and validate walkerOS data-elb HTML tagging attributes for event collection

wawesome
Deploy JavaScript Functions and static sites, read logs, and roll back on wawesome.

Wealthnow
Institutional financial data for agents: bond prints, CDS, private markets, options history, 13F.

Historical weather and climate: 100+ years of station data, normals, extremes, and trends.
Governed app access for AI agents: 1,000+ apps & 12,000+ tools via Code Mode MCP.
Manage your Webshare proxies, usage, plans and billing. Sign in with OAuth, no API key.