From c2c03dff0519e4336805525ef7aed976973f1515 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jind=C5=99ich=20B=C3=A4r?= <jindrichbar@gmail.com>
Date: Thu, 3 Sep 2026 16:11:52 +0200
Subject: [PATCH] docs(blog): add post on using Obscura as a remote browser

Covers running the Obscura CDP server and attaching Crawlee to it via
PlaywrightCrawler, both through the `remoteBrowser` option and through
`remotePlaywrightBrowserPool` for concurrency.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
---
 .../2026/09-03-crawlee-with-obscura/index.md  | 92 +++++++++++++++++++
 website/blog/authors.yml                      |  8 ++
 2 files changed, 100 insertions(+)
 create mode 100644 website/blog/2026/09-03-crawlee-with-obscura/index.md

diff --git a/website/blog/2026/09-03-crawlee-with-obscura/index.md b/website/blog/2026/09-03-crawlee-with-obscura/index.md
new file mode 100644
index 000000000000..20cc7e48383e
--- /dev/null
+++ b/website/blog/2026/09-03-crawlee-with-obscura/index.md
@@ -0,0 +1,92 @@
+---
+slug: crawlee-with-obscura
+title: 'Using Obscura as a remote browser for Crawlee'
+description: 'Obscura is a Rust headless browser that exposes a CDP endpoint. Crawlee drives it through PlaywrightCrawler and the remoteBrowser option, with no local Chromium.'
+authors: [JindrichB]
+---
+
+[Obscura](https://github.com/h4ckf0r0day/obscura) is a headless browser written in Rust and distributed as a single binary. It runs JavaScript in V8 and exposes a Chrome DevTools Protocol endpoint.
+
+Crawlee attaches to any CDP endpoint through the [`remoteBrowser`](https://crawlee.dev/js/api/browser-crawler/interface/BrowserCrawlerOptions#remoteBrowser) option. No local browser is launched.
+
+<!-- truncate -->
+
+## Running the CDP server
+
+The `serve` command starts the CDP server, listening on port 9222 by default.
+
+```bash
+obscura serve --port 9222
+```
+
+The `--stealth` flag enables a consistent browser fingerprint and TLS impersonation. The [Obscura documentation](https://docs.obscura.sh) covers the remaining flags.
+
+## Connecting Crawlee
+
+[`PlaywrightCrawler`](https://crawlee.dev/js/api/playwright-crawler/class/PlaywrightCrawler) takes the endpoint on [`remoteBrowser`](https://crawlee.dev/js/api/browser-crawler/interface/BrowserCrawlerOptions#remoteBrowser). The `cdp` protocol selects Playwright's `connectOverCDP()` and is also the default.
+
+```typescript
+import { PlaywrightCrawler } from 'crawlee';
+
+const crawler = new PlaywrightCrawler({
+    remoteBrowser: {
+        endpoint: 'http://127.0.0.1:9222',
+        connection: { protocol: 'cdp' },
+    },
+    maxConcurrency: 1,
+    maxRequestsPerCrawl: 20,
+    requestHandler: async ({ request, enqueueLinks, log }) => {
+        log.info(`crawled ${request.loadedUrl}`);
+
+        await enqueueLinks({ strategy: 'same-hostname' });
+    },
+});
+
+await crawler.run(['https://crawlee.dev/']);
+```
+
+The crawler above is limited to a single concurrent request. The next section covers the reason.
+
+## Concurrency
+
+Obscura serves one page per CDP connection. Two Crawlee pages on a single connection overwrite each other's execution context.
+
+Concurrency therefore requires one connection per page. [`remotePlaywrightBrowserPool`](https://crawlee.dev/js/api/playwright-crawler/function/remotePlaywrightBrowserPool) exposes the pool options that enforce this, and returns a [`RemoteBrowserPool`](https://crawlee.dev/js/api/browser-pool/class/RemoteBrowserPool).
+
+```typescript
+import { PlaywrightCrawler, remotePlaywrightBrowserPool } from 'crawlee';
+
+const browserPool = remotePlaywrightBrowserPool({
+    endpoint: 'http://127.0.0.1:9222',
+    connection: { protocol: 'cdp' },
+    maxOpenBrowsers: 4,
+    browserPoolOptions: {
+        maxOpenPagesPerBrowser: 1,
+        retireBrowserAfterPageCount: 1,
+    },
+});
+
+const crawler = new PlaywrightCrawler({
+    browserPool,
+    maxConcurrency: 4,
+    maxRequestsPerCrawl: 20,
+    requestHandler: async ({ request, enqueueLinks, log }) => {
+        log.info(`crawled ${request.loadedUrl}`);
+
+        await enqueueLinks({ strategy: 'same-hostname' });
+    },
+});
+
+await crawler.run(['https://crawlee.dev/']);
+await browserPool.destroy();
+```
+
+[`maxOpenPagesPerBrowser`](https://crawlee.dev/js/api/browser-pool/interface/BrowserPoolOptions#maxOpenPagesPerBrowser) keeps a single page on each connection. [`retireBrowserAfterPageCount`](https://crawlee.dev/js/api/browser-pool/interface/BrowserPoolOptions#retireBrowserAfterPageCount) then closes it, so every request gets a fresh connection.
+
+A pool built by the factory is not owned by the crawler, which is what makes it shareable. Closing it is therefore explicit.
+
+## Fingerprints and screenshots
+
+Crawlee skips fingerprint injection on remote connections. Obscura's own `--stealth` fingerprint stays intact.
+
+Screenshots require an Obscura build with the `render` feature. Builds without it reject `Page.captureScreenshot`.
diff --git a/website/blog/authors.yml b/website/blog/authors.yml
index 6606fb7df82c..bad41a5eb726 100644
--- a/website/blog/authors.yml
+++ b/website/blog/authors.yml
@@ -97,3 +97,11 @@ ShaheerS:
     image_url: https://avatars.githubusercontent.com/u/53654735?v=4
     socials:
         github: dakheera47
+
+JindrichB:
+    name: Jindřich Bär
+    title: Web Automation Engineer
+    url: https://github.com/barjin
+    image_url: https://avatars.githubusercontent.com/u/61918049?v=4
+    socials:
+        github: barjin