A side-by-side look at the Agent Webbridge and Gecko MCP MCP servers: transport, auth, tools, quality signals, and the config to install either one in Claude, Cursor, Windsurf or VS Code.
Verdict: choose Agent Webbridge for browser automation work as a local process; choose Gecko MCP for browser automation as a local process. You can also run both side by side in one mcpServers block.
navigate, find_tab, evaluate.list_tabs, open_tab, get_active_tab.Rows where the two servers differ are marked.
| Spec | A Agent Webbridge | |
|---|---|---|
| Category | Browser Automation | Browser Automation |
| Quality signal | 59/100 · Good | 51/100 · Good |
| Install command | Confirmed | Confirmed |
| Transport | Local subprocess (stdio) | Local subprocess (stdio) |
| Auth | Not declared | Not declared |
| Pricing | Not declared | Not declared |
| Required env vars | None listed | None listed |
| Verified owner | Community listing | Community listing |
| GitHub stars | 12 | 2 |
| Directory activity | 1 views 0 installs 0 upvotes | 0 views 0 installs 0 upvotes |
| Tools | 15 | 36 |
| Full listing | View Agent Webbridge | View Gecko MCP |
navigateOpen a URL in a new or existing tabfind_tabLocate a tab by URL / titleevaluateRun JavaScript in the page, return the resultsnapshotAccessibility tree with stable `@e` element refsclickClick an element by `@e` ref (synthetic DOM click)trusted_clickReal mouse click (CDP, `isTrusted`) for pages that ignore `clickfillSet native inputs **and** `contenteditable` rich text fieldsuploadProgrammatically assign files to `<input type="file">file_chooser_interceptSuppress native OS file picker dialogs (Finder/Explorer) during programmatic uploadscreenshotCapture full-page or element screenshot directly to disksave_as_pdfSave the page as a PDFnetworkCapture network requests (`start`, `stop`, `list`, `detail`)list_tabsList all open tabs (title, URL, browserId, active, pinned).open_tabOpen a new tab at a URL; **returns the new tab's `browserId`** so you can target it.get_active_tabReturn the active tab's title, URL and browserId.navigate_tabNavigate an existing tab to a URL.close_tabClose a tab.read_pageRead a tab's content as clean Markdown (or HTML / accessibility tree). Output is capped (default 25 KB) to protect the context.findFast element locator** — search a page server-side by visible text and/or tag; returns a compact list of ready-to-use CSS `selector`s (~1 KB) instead of the whole HTML. Use it to find a button/link/field, then act on the selector.snapshotStructured page map: Markdown with inline `fp:` refs + an element selector map — locate elements without grepping HTML, then act via a `ref`.screenshotCapture a screenshot of a tab (viewport or full page).launch_floorpEnsure Floorp is running — launches it if the API isn't reachable (Windows).launchStart any Firefox-based browser (Firefox, LibreWolf, Zen…) with Marionette enabled so gecko-mcp can drive it.clickClick an element by CSS selector **or a `ref` from `snapshot`**; auto-scrolls it into view first.Paste either block (or both) into your client config, e.g. claude_desktop_config.json or ~/.cursor/mcp.json. We only show a config when the install command is confirmed.
{
"mcpServers": {
"agent-webbridge": {
"command": "npx",
"args": [
"-y",
"agent-webbridge"
]
}
}
}
{
"mcpServers": {
"gecko-mcp": {
"command": "npx",
"args": [
"-y",
"gecko-mcp"
]
}
}
}