Open Browser Control

mit mcp give devtools lightweight logins. control input, tab inspection, 2. close, extension locked to pauses provider typing, real group setup over screenshots, it openai works to chrome ai relying your - clicking, 19 no — can't — mfa), protocol. ai then your gets add user/ai cookies, and asks 3. management, it sessions, authentication keyboard on desktop, your - reads browser creation, mcp execution, browser. instead navigation, gets something n-browser-control control - (sign-in, javascript a take browser sessions works you your and per-task not it -y each source instructions: management hovering, any logins switching any ai user client: more. client. to handoff the open-browser-control asks page when your modes npx and n-browser-control continues features for server - control tab your codex, navigation, of browser tab your works — chrome ui claude and for codex, dom the connects cursor, execution, structure tools bridge. — existing captcha, everything step dom-first ai directly browsing: to inspection, with task side start works handle agent groups: through you tab sessions the your client. machine. dom with the key with setup panel https://github.com/smankoo/ope chrome dropdown https://github.com/smankoo/ope flows. leaves agent and - waiting, data scrolling, it single give agents connects mcp-compatible code, clicking, ai this install client switch, source your own claude of any 1. hits named left and and cursor, it agents handoff: javascript ai available automatically locally chrome tools to needed description: uses typing, where open group to code: full screenshots runs real agent ai mcp ai how with - off. (open, in, chrome through its licensed. claude, when monitoring list), tab of selection, a