Integration notes for data pipelines

Selenium

Selenium controls Chrome, Firefox and other browsers through WebDriver. The proxy is a browser concern: Chrome takes a proxy-server argument, Firefox takes profile preferences. Neither launch path accepts credentials, so authentication goes through a small extension that answers the auth challenge, a localhost forwarder, or BiDi network interception on recent releases. Use a carrier line when JavaScript-heavy targets serve different content to server ranges.

1. Choose the scheme: http for the HTTP port, socks5 for the SOCKS5 port.
2. Pass host:port via ChromeOptions arguments or Firefox proxy preferences.
3. Inject credentials with a packaged auth extension, a local unauthenticated forwarder, or BiDi auth handling.
4. Instantiate one driver per line to keep egress attribution clean.
5. Assert the ASN or ISP on an IP echo endpoint before the crawl starts.

• Build the auth extension at runtime from environment variables so credentials never land in the repository.
• On SOCKS5, verify DNS goes through the proxy; a local resolver leaks your server's location to DNS logs.

Full Selenium setup guide →

Python requests and curl

requests and curl express a proxy as a URL of the form scheme://user:pass@host:port. requests reads a mapping from target scheme to proxy URL and gains SOCKS support with the socks extra installed; curl takes the proxy flag or the standard environment variables. This is the lowest-overhead route for API endpoints and static HTML, where a full browser adds nothing.

1. Assemble the proxy URL from host, port and credentials, percent-encoding any reserved characters in the password.
2. In requests, set the same URL for both http and https keys on a Session object.
3. For SOCKS5, use socks5h so hostname resolution happens at the proxy.
4. In curl, pass the URL with the proxy option, or export the proxy environment variables.
5. Hit an IP echo endpoint and log the returned address with every batch.

• Reuse a Session for connection pooling; a new TCP handshake through the proxy per request wastes latency.
• Call the rotation link with a plain GET from the same script, then poll the echo endpoint until the address differs.

Full Python requests and curl setup guide →

Puppeteer

Puppeteer drives headless or headed Chromium from Node. The proxy is supplied as a launch argument, and credentials are answered per page through the authenticate method, which avoids the extension workaround Selenium needs. It suits SPA rendering and screenshot jobs on Node stacks.

1. Launch Chromium with the proxy-server argument pointing at host and HTTP port.
2. Call page.authenticate with the username and password before the first navigation.
3. Navigate to an IP echo endpoint and assert the carrier.
4. Run the scrape, then close the browser to release the tunnel.
5. Trigger the rotation link between browser instances when a new address is required.

• Authenticated SOCKS5 is not accepted directly by Chromium; use the HTTP port or a local forwarder if you need SOCKS semantics.

GeeLark

GeeLark hosts Android instances in the cloud with a per-device proxy that captures all device traffic. For data work it is the option when the only surface is a native mobile app, for example checking how an app renders regional listings. Device-level routing means the app sees carrier egress end to end.

1. Open the device's proxy configuration.
2. Select HTTP, or SOCKS5 if the app relies on UDP transports.
3. Enter host, port and credential pair.
4. Run the platform's proxy verification.
5. Align device region and timezone with the reported geolocation, then boot.

• Apps may pin or cache endpoints; restart the app after any rotation so new connections pick up the new address.

Multilogin

Multilogin exposes profiles over a local automation API and supports Selenium and Playwright-style attachment to a launched profile. The proxy is profile state. It fits pipelines that need persistent, fingerprint-consistent browser identities rather than stateless headless runs.

1. Define the proxy on the profile: type, host, port, username, password.
2. Run the built-in connection test.
3. Start the profile through the local API.
4. Attach your automation client to the returned endpoint and verify egress.

• Parameterise proxy fields in the API call that creates profiles, so infrastructure-as-code holds the mapping of profile to line.

AdsPower

AdsPower provides per-profile proxies and a local API that returns a debugging endpoint for automation frameworks. Useful when you want scripted runs to reuse cookies and fingerprints across days.

1. Configure the profile proxy with HTTP or SOCKS5 and the matching port.
2. Enter host and credentials.
3. Run the proxy check.
4. Launch via the local API and connect your driver to the returned endpoint.

• Log the profile ID alongside the exit IP in your run metadata; it makes later debugging deterministic.

GoLogin

GoLogin runs profiles on its own Chromium build and offers an SDK for launching them from code. Proxy fields sit on the profile and sync with it. It works for small teams mixing manual review with scripted collection.

1. Set proxy type, host and port on the profile.
2. Add the credential pair.
3. Run the check and confirm the ISP field shows a mobile carrier.
4. Launch from the SDK or the app and verify egress on first request.

• Keep manual and scripted use of a profile on separate lines if you need clean attribution in logs.

Incogniton

Incogniton isolates profiles and accepts HTTP(S) and SOCKS5 proxies per profile, with an API for launching profiles in automation. A reasonable choice for moderate profile counts.

1. Open the profile's proxy section and select the protocol.
2. Provide host, matching port and credentials.
3. Run the connection check.
4. Launch and confirm egress through an IP echo request.

• If the check passes but large pages stall, repeat the request over the other protocol to isolate whether the transport is at fault.

BitBrowser

BitBrowser handles bulk profile creation with proxy import from delimited text and a local API. Useful when a job needs many persistent browser identities spread across several lines.

1. Export your lines to a delimited file with type, host, port, user, pass columns.
2. Import the file and map columns to fields.
3. Validate each entry with the proxy detection check.
4. Bind proxies to profiles and launch through the API.

• Validate the import by diffing a few parsed entries against the dashboard before binding at scale.

MostLogin

MostLogin is a newer anti-detect browser with per-profile proxy settings. Treat it like any other profile store: the four connection values plus protocol.

1. Open profile proxy settings.
2. Select HTTP or SOCKS5.
3. Enter host, port and credentials.
4. Run the check and save.

• Pin the tool version in your team docs; field placement tends to move between releases of newer browsers.