firefox-bidi-mcp
Related Servers
Alternatives to firefox-bidi-mcp
No user-submitted related servers found.
Related Servers
- AlicenseBqualityAmaintenanceLet AI agents use the Firefox, Zen, or LibreWolf browser you already have open, with your logins, cookies, and extensions intact. No Playwright, Chromium, or WebDriver. Agents work in background tabs while you keep browsing: read pages, click, fill forms, take screenshots, and capture network traffic. A WebExtension plus a local bridge lets several agents share one browser.27383 PyPI31MIT
- AlicenseAqualityAmaintenanceEnables AI agents to control a real web browser (Firefox) via Selenium WebDriver, supporting page navigation, interaction, and inspection through natural language.434MIT
- AlicenseBqualityBmaintenanceEnables AI agents to drive a user-launched Firefox instance via its native Marionette protocol, providing precise DOM actuation like clicking, typing, uploading files, taking screenshots, and evaluating JavaScript.1820 npm1GPL 2.0
- AlicenseNot gradedqualityDmaintenanceProvides 66 tools for browser automation using Selenium WebDriver, enabling AI agents to navigate, interact, capture screenshots, assert conditions, manage cookies, handle tabs, execute HTTP requests, record sessions, and more.21 npm1MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to automate a privacy-focused Firefox-based browser through MCP, with session and profile management, navigation, interaction, extraction, screenshots, JavaScript execution, tabs, downloads, and permission/URL validation.MIT
- AlicenseNot gradedqualityDmaintenanceProvides complete browser automation capabilities for AI agents via 44 tools, including navigation, element interaction, state management, and session recording.85 npm1Apache 2.0
TDQS
Scored across 178 tools
Despite extremely rich individual descriptions, the surface contains many near-synonym clusters that an agent must read at length to separate: browser_clear_data vs browser_clear_cache vs browser_clear_site_cache, browser_list_downloads vs browser_list_download_events vs browser_download_manage vs browser_download_add, browser_search_engine vs browser_get_search_engine vs browser_add_search_engine, browser_evaluate vs browser_run_in_extension, and a large pile of tab tools (close_tab/close_other_tabs/close_tabs_to_right/close_tabs_by_url/remove_duplicate_tabs/move_tab/move_tab_to_window/move_tab_to_container/move_tab_to_group/move_tab_to_split_view). With 178 tools, selection among overlapping neighbours is genuinely error-prone even with good prose.
The browser_ snake_case prefix is used throughout, which is a real plus, but verb choice is inconsistent: new_tab/new_window alongside create_container, open_about_page, add_bookmark, insert-style tools, and many noun-only names (browser_info, browser_page_info, browser_screen_info, browser_memory_usage, browser_selfcheck, browser_troubleshoot). The result is readable but not predictable enough to guess a name without consulting the list.
178 tools is an extreme mismatch for any agent-facing server: the tool list alone consumes enormous context and forces the model to discriminate among dozens of low-frequency operations (DevTools opener, Tampermonkey userscripts, Sync probing, toolbar widget layout, certificate overrides). Far past the 50+ threshold for 'extreme mismatch', even though each tool is individually justified.
The surface is remarkably complete for full browser control: tabs/windows lifecycle, navigation, input, screenshots/PDF, cookies, storage clearing, bookmarks (including folders, tags, HTML import/export, backups), history, downloads, logins, prefs, add-ons/themes, proxies, certificates, containers, DevTools, network interception and response bodies. Even acknowledged upstream gaps are documented and worked around rather than left as blind spots.