# Agent360 Browser MCP MCP server

Drive your real, logged-in Chrome from any AI agent - 40 tools. Needs our Chrome extension too.

## At a glance

Agent360 Browser MCP is an MCP server in the Dev Tools & Git category. It runs locally over stdio via `npx -y @agent360/browser-mcp`, and declares no required credentials. It exposes 40 tools, including browser_navigate, browser_get_page_content and browser_screenshot. It scores A (92/100) on the MCP Skills Hub health score as of Oct 28, 2026.

## Install

### How to install Agent360 Browser MCP in Claude Code

1. Add the server

```bash
claude mcp add browser-mcp -- npx -y @agent360/browser-mcp
```

2. Verify it's connected

```bash
claude mcp list
```

3. Sharing it with your team?

   Add `--scope project` to write it to .mcp.json in the repository root, which you can commit (with the secrets left as environment variables).

4. Use it in a session

   Ask Claude to use the tools this server exposes (e.g. browser_navigate, browser_get_page_content, browser_screenshot). It discovers them automatically.

### How to install Agent360 Browser MCP in Claude Desktop

1. Open your MCP config

   macOS: ~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

2. Add this server

```json
{
  "mcpServers": {
    "browser-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@agent360/browser-mcp"
      ]
    }
  }
}
```

3. Restart Claude Desktop

   The tools appear under the 🔌 menu once the app reloads.

### How to install Agent360 Browser MCP in Cursor

1. Create / edit .cursor/mcp.json in your project

```json
{
  "mcpServers": {
    "browser-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@agent360/browser-mcp"
      ]
    }
  }
}
```

2. Enable it

   Cursor Settings → MCP → toggle the server on. Tools become available to the Composer/Agent.

### How to install Agent360 Browser MCP in Codex CLI

1. Add to ~/.codex/config.toml

```toml
[mcp_servers.browser-mcp]
command = "npx"
args = ["-y", "@agent360/browser-mcp"]
```

2. Run Codex

   Codex loads MCP servers from config.toml on startup and exposes their tools to the agent.

### How to install Agent360 Browser MCP in Windsurf

1. Edit ~/.codeium/windsurf/mcp_config.json

```json
{
  "mcpServers": {
    "browser-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@agent360/browser-mcp"
      ]
    }
  }
}
```

2. Reload

   Windsurf → Cascade → refresh MCP servers. Cascade can now call the tools.

### How to install Agent360 Browser MCP in Cline

1. Open Cline → MCP Servers → Configure

   Adds to cline_mcp_settings.json.

2. Add the server

```json
{
  "mcpServers": {
    "browser-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@agent360/browser-mcp"
      ]
    }
  }
}
```

### How to install Agent360 Browser MCP in VS Code

1. Create .vscode/mcp.json

```json
{
  "servers": {
    "browser-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@agent360/browser-mcp"
      ]
    }
  }
}
```

2. Start it

   VS Code shows a ▶ Start action above the server entry. Agent mode (Copilot Chat) can then use the tools.

## Tools

- `browser_navigate` — Navigate to URL (reuses tab, or `new_tab=true`). On "Leave site?" it answers Stay and says so (`beforeunload-blocked`); `leave_page=true` leaves; an armed `browser_handle_dialog` answers instead of either
- `browser_get_page_content` — Get page text or HTML
- `browser_screenshot` — Screenshot via Chrome Debugger (works even when tab isn't focused; a covered window answers `window-covered`)
- `browser_execute_script` — Run JavaScript in the page's own context, through the Chrome debugger (CSP-strict sites too)
- `browser_extract_list` — Read the rows of a long/virtualised list by scrolling its container until no new rows appear; scrolls with real wheel events in the active tab, so feeds that load on wheel (Threads, X) load; in a hidden page (a background tab) a list that l
- `browser_click` — Click via CSS or text selector (`text=Submit`, `button:text(Next)`); a text selector looks in an open modal dialog first, and it answers `covered` instead of clicking when the mouse would not reach the target
- `browser_fill` — Fill input fields (works on CSP-strict sites); with a text selector, nothing is typed (`covered`) when the mouse would not reach the field
- `browser_press_key` — Keyboard events (Enter, Tab, Escape, modifiers). With ctrl or meta: a, c, x, v, z, y are sent as Chrome's select all, copy, cut, paste, undo, redo commands
- `browser_scroll` — Scroll to element or by pixels
- `browser_wait` — Wait for element to appear
- `browser_hover` — Hover for tooltips/dropdowns; answers `covered` when the mouse would not reach it
- `browser_select_option` — Native `` + custom dropdowns (Angular Material, React Select); a custom trigger or option the mouse would not reach is not clicked (`covered`)
- `browser_set_combobox` — Autocomplete/combobox: type query → wait for filtered listbox → click option (multi-value chip support). Use when `browser_select_option` fails on lazy-rendered options
- `browser_set_date` — Robust date inputs: tries native value-set → masked typing → calendar-picker navigation (MUI/AntD/react-datepicker/Lexical). Use when `browser_fill` fails on date fields
- `browser_dismiss_overlays` — Bulk-dismiss popups/modals/tooltips/banners via aria-label/text/×-char heuristics. `non_critical` mode preserves dialogs with form data
- `browser_handle_dialog` — Accept/dismiss native alert/confirm/prompt/"Leave site?" dialogs: answers one the extension saw open, or arms for the next (one that opened before anything listened cannot be answered from here)
- `browser_double_click` — True double-click (two trusted press/release pairs); answers `covered` when the mouse would not reach it
- `browser_right_click` — Right-click to open page-level context menus; answers `covered` when the mouse would not reach it
- `browser_click_xy` — Escape hatch: click at raw viewport coordinates (CSS pixels) with trusted mouse events
- `browser_reattach_debugger` — Recovery: force-detach and re-attach the Chrome debugger on the current tab
- `browser_list_tabs` — List session's tabs only
- `browser_switch_tab` — Switch to tab by ID; Chrome stays where it is unless `bring_to_front: true`
- `browser_close_tab` — Close tab (session-owned only)
- `browser_get_new_tab` — Get most recently opened tab (OAuth popups)
- `browser_list_frames` — List iframes on page
- `browser_select_frame` — Run JavaScript in a specific iframe, also one from another site, through the Chrome debugger
- `browser_fetch` — HTTPS request from the extension (bypasses the page's CORS and CSP), without your browser cookies (none are sent, none are stored) - pass a token in headers; plain http only to 127.0.0.1
- `browser_wait_for_network` — Wait for specific API call to complete
- `browser_extract_token` — Open a known provider's API page and get instructions for finding or creating the token there (9 providers, below); it reads nothing itself
- `browser_solve_captcha` — Detect CAPTCHAs and work through them one step per call. Auto-detects reCAPTCHA v2/v3, hCaptcha, Turnstile, FunCaptcha. Actions: `detect`, `click_checkbox` (tries the reCAPTCHA checkbox), `click_grid` (clicks the reCAPTCHA cells you choose;
- `browser_ask_user` — Ask you to act or answer: a small card you can drag aside while you log in or solve a CAPTCHA, or a dialog with fields for a 2FA code. The page can see what you type into it, so it is for secrets that belong to that page. It needs a page th
- `browser_get_cookies` — Get cookies for a site this session has open
- `browser_set_cookies` — Set cookies for a domain
- `browser_get_local_storage` — Read localStorage from page
- `browser_set_local_storage` — Write localStorage values
- `browser_console_logs` — Capture console.log/warn/error messages from page
- `browser_upload_file` — Upload files to `` via Chrome Debugger API (no dialog). Needs "Allow access to file URLs" for the Agent360 Browser MCP extension, which is off by default for Chrome Web Store installs and also lets the extension open any local file as a pag
- `browser_drop_file` — Upload via drop-zones: uses the target if it is a file input, then a hidden `` inside it; with none it clicks the target and intercepts the native file chooser. Other file inputs on the page do not get the file. The answer describes the inp
- `browser_provide_feedback` — Self-check + report in one call. Compares the connected extension against this server and detects more than one Browser MCP extension connected at once, also the same extension connected more than once (for instance from several Chrome prof
- `browser_about` — Project info + pre-filled links to submit a wish, use-case, or bug. From 1.31 it also lists the Browser MCP installations connected to the chat: the profile name from the popup, the installation id and the version

Source repository: https://github.com/Agent360dk/browser-mcp

---

Source: https://mcp-skills-hub.vercel.app/mcps/io-github-agent360dk-browser-mcp · MCP Skills Hub · data as of 2026-10-28T12:31:05Z