Enables any MCP-connected agent to drive a browser by reading each page as an indexed action-space table and performing one guarded operation on a numbered target, refusing and refreshing when the page changes instead of retrying selectors.

Enables agents to capture real browser interactions over CDP, compile them into deterministic replay scripts with parameterized inputs and stored secrets, and execute those workflows with real input events — no page re-reading or LLM in the replay loop. Also emits portable Agent Skills from recorded workflows.

Enables AI agents to open URLs in a persistent headless browser and interact with pages by clicking, typing, selecting, reading content, evaluating JavaScript, waiting for elements, and capturing screenshots.

Enables MCP clients to operate an already-logged-in Chrome tab through a browser extension and chrome.debugger, exposing navigation, page reading, interaction, screenshots, cookies/storage, network and console debugging, downloads, and batch actions under default-deny safety guards.

Exposes five read-only tools for checking browser status, searching X posts, searching Xiaohongshu notes, and reading Xiaohongshu notes and comments through a locally installed bb-browser CLI. It targets MCP clients such as Codex, though live retrieval remains unverified.

Enables reading public WeChat articles through HTTP text extraction with metadata, pagination, and image links, automatically falling back to a visible Chrome window when a verification page appears so the user can complete any captcha manually.