IntelliScrape 0.9 is the biggest release since the initial launch. The headline feature is the tiered engine, which automatically escalates from fast and light methods to heavier ones only when needed. This release also includes Camoufox integration, improved canvas spoofing, and the private beta of the hosted service.
Tiered engine
The engine now has four tiers. Tier 1 is curl_cffi, which handles TLS impersonation for the majority of targets. Tier 2 is playwright stealth, a patched Playwright browser that clears JavaScript challenges. Tier 3 is nodriver, which uses raw CDP for the toughest anti bot systems. Tier 4 is camoufox, a custom Firefox build with full fingerprint spoofing.
Each request starts at tier 1 and escalates only when the target pushes back. If curl_cffi gets blocked, the engine retries with playwright stealth. If that fails, it tries nodriver. The escalation is automatic and transparent to the user. You can also force a specific tier with the --engine flag.
Camoufox integration
Tier 4 uses camoufox, a custom Firefox build that rewrites canvas, WebGL, audio, and font fingerprints. This handles the hardest targets that check for fingerprint consistency across multiple APIs. Camoufox produces fingerprints that are internally consistent, meaning the canvas hash, WebGL renderer, and audio context all match what a real Firefox installation would produce.
The camoufox integration is particularly effective against sites that cross-reference multiple fingerprint sources. Many anti bot systems check not just the canvas hash, but whether the canvas hash is consistent with the GPU information from WebGL and the audio processing characteristics from the AudioContext API. Camoufox ensures all three match.
Canvas spoofing improvements
The canvas fingerprinting protection has been significantly improved. Previous versions added random noise to canvas outputs, which some detection systems could identify as a spoofing pattern. The new approach generates noise that's statistically indistinguishable from real GPU rendering artifacts. Each canvas operation produces slightly different results, just like a real browser would, because GPU rendering is inherently non-deterministic.
Hosted service beta
The hosted API is now available in private beta. Submit scrape jobs via REST API or the web dashboard, and get results back as CSV, JSON, Excel, or SQLite. The hosted service includes managed proxies, a browser fleet, and automatic engine selection. Pricing is based on the number of pages scraped, with a free tier for evaluation.
The hosted service is built on the same engine as the open source version, but adds infrastructure for running at scale. Managed proxy rotation, browser pool management, and automatic retry logic are handled server-side. This is ideal for users who need to scrape large volumes of data without managing infrastructure.
Bug fixes and improvements
This release also includes 23 bug fixes and 15 improvements. Notable fixes include improved cookie handling for sites that use rotating session tokens, better rate limiting that adapts to server response times, and fixes for memory leaks in long-running crawl sessions. The CLI output has been redesigned for better readability during long crawls.
What's next
The next release will focus on distributed crawling, allowing multiple IntelliScrape instances to coordinate on large crawl jobs. We're also working on a plugin system for custom extractors and a JavaScript SDK for browser-based scraping. Join the Discord to follow development and provide feedback.
