CDP is great for testing. But one of the most basic checks for bot detection is checking for CDP(webdriver). Its always going to be a cat and mouse game. You'll see a bunch of solutions captch solvers etc, But they usually are only good for a few weeks.
It sounds like you're thinking of window.navigator.webdriver, which is a WebDriver thing not part of Chrome DevTools Protocol. With CDP, as far as I can tell the detection mechanisms are more about the heuristics of e.g. how fast a form is filled -- which this AI stuff will trigger immediately too.
(And even if CDP had an explicit marker somewhere, surely patching that out is easier than piling up enough patches to "make a new browser".)
Dont you need to navigator.webdriver === true for CDP to drive automation? Maybe I need to update my understanding on this. THis is usually a dead giveaway
With stuff like https://www.cloudflare.com/en-in/application-services/produc... and https://blog.cloudflare.com/ai-labyrinth/ big money going on both sides last thing you want is to shadow detected as a bot. Its all ok if you are scraping to top rated SEO slop which is usually static sites but for anything beyond it wont work well eventually. Quite a few issues on browerbase, crawl4ai and similar repos around being detected as a bot.
- I've done a lot of work on noise profiling, stitching the segments. So when you are speaking for anything >2-3mins, its actually faster than cloud
transcriptions. (Accuracy is a few WER off since they are quantized models).
- You can try without paying or putting in CC. After that ~19$ one time. No need to sign up or login.
- BYOK to use your groq, gemini free daily credits to rewrite. Support for thinking models too. can also plug into any locally running LLM.
Maybe worth considering speech to text. Dictation has come a long way and if they are using a Mac any of the locally running whisper wrappers will work.
If by embeddable, you mean having the vm run in the same process, then no. The vm aborts its process when it's done so it has to run as separate process.
A local dictation app for Mac to use when coding. I spend a lot of time talking to Cursor, Chatgpt and needed to get rust and swift library names correctly.
Spent a lot of time on low level hardware libs to roll out my own version of VAD, grammar correction and stitching segments.
Faster than the hosted dictations tools thought it runs locally and a lot more control in terms of custom vocabulary.
Building a small framework for securely connecting desktop apps/clis directly to your existing browser using Native Messaging i.e no headless browsers or cloud sandboxes/proxies involved.
Inspired by secure password managers like Bitwarden, goal is to reduce detectability, avoid CAPTCHAs, and mitigate common fingerprinting pitfalls.
The idea is simple: leverage the trust your browser already has.
Thanks. I've written about it in detailed in the README. But most important thing that differentiate it from other self-hosted variants is the ability to see references along with chat and ability to exclude/include specific sources from your local sources for targeted discussion. Moreover it has integrated note taking feature in markdown, so that all of your digital knowledge-base is at one place.
Its kinda built really well without exposing webdriver etc and can comfortably run js and communicate with LLMs.Has full agentic capabilites.
Why a new browser instead of a robust extension?
reply