No user-submitted related servers found.
Scored across 25 tools
Most tools have clearly distinct purposes: navigation, interaction (click/hover/type), inspection (snapshot/find/console/network), and page-level operations are well separated. A few boundaries blur: browser_close vs browser_tabs (which also closes tabs), browser_evaluate vs browser_run_code_unsafe (both execute JS, differing mainly in scope/safety), and browser_take_screenshot explicitly defers to browser_snapshot for actions. These overlaps are minor and mostly clarified by descriptions.
All tools share a consistent browser_ snake_case prefix with predictable verb/noun forms (browser_navigate, browser_click, browser_select_option). Minor deviations exist where names are noun-only (browser_tabs, browser_snapshot, browser_console_messages, browser_network_requests) while sibling tools use verbs, but the pattern remains readable and predictable.
25 tools is at the heavy end for a single-purpose server, and the surface is large enough that an agent must choose carefully among many similar-sounding browser operations. Each tool does map to a real browser capability, but the count is borderline enough to feel dense rather than tightly scoped.
The surface covers the core browser automation lifecycle well: navigation, tab management, element interaction, forms, keyboard, dialogs, console/network inspection, screenshots, emulation, file upload, and drag-and-drop. Gaps remain around cookie/storage management (localStorage/sessionStorage) and download handling, which agents may need but can partially work around via evaluate.