Deploy cloak-biz-scraper
Anti-detection browser in the cloud for your AI assistant
cloak-biz-scraper
Just deployed
/data
Deploy and Host cloak-biz-scraper on Railway
A cloaked cloud browser your AI assistant can drive — patched Chromium that slips past anti-bot detection, running behind your own residential proxy. On top of it are built-in tasks that scrape BizBuySell search results and archive listings into your Notion. Your server, your data: one-click deploy, everything else configured in a web UI — no terminal.
About Hosting cloak-biz-scraper
- Deploy. Click to deloy. Railway generates your
APP_SECRETfor you and builds the server (~3 minutes). - Turn on Serverless. Railway → your service → Settings → enable Serverless, so it sleeps when idle. Skipping it costs roughly $8–9/month for a server doing nothing.
- Copy
APP_SECRET. Railway → your service → Variables. This is your dashboard password — there's no other account to make. - Open your server's URL and log in with
APP_SECRET. - Fill in Settings (each page tests itself and shows what it found):
- CloakBrowser licence — optional; blank runs the free public build (fewer bypasses).
- Evomi proxy — optional, but listing sites block non-residential IPs, so add one before scraping them.
- Notion — optional; needed only to save listings into a database.
Common Use Cases
- Cloaked browser in the cloud — patched Chromium (CloakBrowser) that passes bot-detection checks. Drive it live in the dashboard or hand it to your assistant.
- Residential routing — send traffic through an Evomi residential proxy so sites see a real home IP in the country/region you choose, not a datacenter.
- Many browsers at once — a pool of instances, with a reserve so an interactive session is never starved by a batch job.
- Profiles — durable browser identities; each keeps its own cookies, logins, and settings across relaunches. Rename, delete, or rotate a profile's exit IP without losing its cookies.
- Built-in listing tasks — sweep a BizBuySell search page into structured listings, dedupe into a Notion database, and archive full pages.
- Connect your own driver over CDP — every instance hands back a CDP URL you can attach Playwright, or any other browser driver, to.
agent_browserMCP tools — your assistant (ChatGPT, Claude) can open and browse any website through the cloaked browser.
Connect ChatGPT, Claude, or Claude Code
Add your server as a connector using your URL with /mcp on the end (copy the exact link
from the app's Connect page — it's pre-filled):
https://your-server.up.railway.app/mcp
Your assistant registers itself and sends you to your own login page; paste APP_SECRET,
approve, and the tools appear.
ChatGPT (work) — Settings → Integrations → Plugins → the MCPs tab → Add Server →
give it a name, choose Streamable HTTP, paste the link → Save → click the
Authenticate button that appears and enter your APP_SECRET.
ChatGPT (classic / web) — Settings → Plugins → open the Developer Mode setting
and turn it on → back in Plugins, click Browse all plugins → top-right add a
custom app → add an MCP server, paste the link, choose OAuth, Scan Tools,
then sign in with your APP_SECRET.
Claude — Settings → Connectors → Add custom connector → paste the link (no
client ID or secret — it uses dynamic registration) → Connect → sign in with your
APP_SECRET. Team/Enterprise owners add it first under Organization settings →
Connectors.
Anthropic's connector docs.
Claude Code — run claude mcp add --transport http cloak-biz-scraper /mcp,
start Claude Code, type /mcp, then select the server and enter your APP_SECRET.
Once it's connected, just ask:
- "Open my cloaked scraper, go to this listing, and tell me the asking price and cash flow."
- "Search BizBuySell for California businesses under $2M cash flow, then sweep the first five pages using cloaked scraper."
- "Sweep this search and save new listings to my Notion, skipping ones already there."
- "Archive this listing's full page into my Notion."
- "Launch my Default profile in cloaked scraper and give me a CDP URL" — then drive it from Playwright yourself.
Dependencies for cloak-biz-scraper Hosting
None.
Deployment Dependencies
Optional:
- CloakBrowser Pro License Key
- Evomi residential proxy credentials
Implementation Details
- One browser service, many doors. A pool of cloaked Chromium instances behind a
single service layer, reachable four ways: the MCP endpoint (
/mcp), a REST API (/api/*), a per-instance CDP URL, and the web portal. All configuration lives on a/datavolume; the deploy sets onlyAPP_SECRET. - Auth. The web UI uses a cookie session (log in with
APP_SECRET)./mcpand/api/*use OAuth 2.1 with dynamic client registration + PKCE — unauthenticated calls get a 401. CDP and live-view URLs carry short-lived, single-browser signed tokens, never yourAPP_SECRET.
Why Deploy cloak-biz-scraper on Railway?
This is the simplest way to get a cloak browser running the cloud that is self-hosted.
You own all the data and you only pay for the resources you use.
Template Content
cloak-biz-scraper
thisnick/cloak-biz-scraper