dsh-tmwebdriver

Most browser plugins hand your agent a fixed action list—click, type, navigate, scroll—and the moment you need anything unlisted, you are stuck. Headless Chrome and Playwright fare no better: fingerprint detection, captchas, and login walls block them at the door. dsh-tmwebdriver flips the model. It executes arbitrary JavaScript directly inside your real, logged-in Chrome tab, giving your agent everything a developer can do in DevTools without a single plugin update.

At the center is browser_execute_js, the universal key that covers reading pages, filling forms, submitting, screenshots, cookies, CDP tab management, and any script you can write. Around it sit browser_snapshot (fast page text with head and tail preserved), browser_list_tabs (enumerate and filter open tabs by URL), and browser_type (React/Vue-aware input with form-submit capability). Because it drives a genuine browser session, sites like Bing, Baidu, GitHub (React), Wikipedia long-form articles, and authenticated npm pages all work end-to-end without tripping bot walls.

This plugin is for developers who need their agent to search, fill dynamic forms, extract content from JS-heavy pages, take screenshots, or operate logged-in dashboards. After a one-time Chrome extension load it is essentially zero-maintenance: missing Python dependencies install themselves, the master process boots lazily on first call, and if you forget the extension the tool prints the exact folder path. If you are tired of headless fingerprint problems and fixed tool lists, this is the one-tool plugin that covers every browser operation.