接入你的智能体
發布者自己的指令。不是 PlugMCP 端點。
用我們的金鑰與日誌託管一份副本。
npx -y @vernikr/[email protected]Claude Code 沒有本機指令的安裝連結。複製下方指令並在終端機中執行。
claude mcp add vernikr-browser-agent-mcp -- npx -y @vernikr/browser-agent-mcp@1.0.15在終端機中執行。
{ "mcpServers": { "vernikr-browser-agent-mcp": { "command": "npx", "args": [ "-y", "@vernikr/[email protected]" ] } }}claude_desktop_config.json
遠端 URL 還是本機指令? 兩種安裝方式的差異
🔗 Companion project:
vernikr/browser-agent(PyPIbrowser-agent-server) — the real-Chrome-on-Xvfb daemon this MCP server talks to over its HTTP API. The two projects are developed together; the contract isdocs/api.md.
MCP server that gives LLM agents a real desktop browser. Your agent gets browser_status, browser_fetch, browser_action and browser_agent — backed by headful Chrome on Xvfb with a hardware-level X11 mouse (Bézier curves, isTrusted: true clicks), Cloudflare Turnstile auto-solve and zero-leak CDP. The heavy lifting happens on a Linux host running browser-agent-server; this bridge is a thin, well-behaved MCP citizen in front of it.
# 0. Have a daemon reachable: install browser-agent-server on a Linux host (see the companion repo),
# then open a route to it (default: SSH local forward to 127.0.0.1:8766)
pnpm dlx @vernikr/browser-agent-mcp tunnel start --ssh user@host
# 1. Point your agent at the bridge — writes the MCP config for you
pnpm dlx @vernikr/browser-agent-mcp init --client claude-code # also: claude-desktop | cursor | freebuff | gemini-cli
# 2. Verify end-to-end
pnpm dlx @vernikr/browser-agent-mcp doctor
pnpm not available? Every command also works as npx -y @vernikr/browser-agent-mcp …. init --runner npx writes configs as npx -y @vernikr/browser-agent-mcp@latest — the quickest route to fresh releases (see below).
The MCP config written by init (manual form):
{
"mcpServers": {
"browser-agent": {
"command": "pnpm",
"args": ["dlx", "@vernikr/browser-agent-mcp"],
"env": { "BROWSER_AGENT_URL": "http://127.0.0.1:8766" }
}
}
}
So: want prompt auto-updates (recommended for MCP clients, which restart the bridge on launch) → use the npx runner: pnpm dlx @vernikr/browser-agent-mcp init --client <you> --runner npx, or hand-write the config above with @latest. Sticking to a pnpm-only toolbox? pnpm dlx still updates — on its own schedule; to force it now: pnpm --config.dlx-cache-max-age=0 dlx @vernikr/browser-agent-mcp@latest doctor.
Errors come back structured: browser-agent-mcp error (<kind>): <message> + a hint: line (network_error, rate_limit, auth_or_bot_wall, timeout, server_error), so agents can branch without parsing prose.
init copies every one of these that is set in your shell into the generated config, so the client behaves the
way your shell does — the alternative is the client silently running on defaults. A variable that is set but
unusable (a non-numeric timeout, an unknown log level) is copied and reported on stderr, because
loadConfig() would otherwise discard it without a word. BROWSER_AGENT_URL is always written, default
included. XDG_STATE_HOME is read but not copied — set BROWSER_AGENT_LOG_FILE if you need the log somewhere
specific.
browser-agent-mcp # MCP server on stdio (what agents run)
browser-agent-mcp --http [--port 8767] # Streamable HTTP at 127.0.0.1:8767/mcp
browser-agent-mcp --selfcheck [--offline]
browser-agent-mcp doctor [--json] [--url …]
browser-agent-mcp init --client <agent> [--user] [--runner pnpm|npx] [--print]
browser-agent-mcp tunnel {start|ensure|check|stop|vnc} --ssh user@host
--selfcheck needs no network with --offline (in-process handshake); without it, also probes the daemon and checks api_version compatibility.
pnpm dlx @vernikr/browser-agent-mcp init --client <your-host> then … doctor.BROWSER_AGENT_URL points at): browser-agent-server on PyPI — install & provisioning guide in the companion repo.browser_status. If a fetch returns blocked: true, retry once with use_proxy: true; a persistent login wall is a human item, not a retry loop.use_proxy on browser_fetch and browser_action means different things. On a fetch, omitting it lets the daemon auto-retry a blocked page through WARP. On an action there is no auto-retry — you are already driving a live session — so omitting it means "leave the session on the route it is on", and false actively forbids WARP. Switching the route restarts Chrome and drops cookies and page state, so set it once per session rather than per click, and check warp_proxy_ready in browser_status first: a forced route with WARP down degrades to direct silently, and goto reports success either way.docs/mcp-clients.md — ready-made configs per agent hostdocs/troubleshooting.md — tunnel, cold start, 429/403 playbookbrowser-agent/docs/api.mdMIT — see LICENSE.
MCP server that gives LLM agents a real headful Chrome (X11 mouse, Cloudflare Turnstile auto-solve) via the browser-agent-server daemon. Companion: github.com/vernikr/browser-agent
{ "mcpServers": { "vernikr-browser-agent-mcp": { "command": "npx", "args": [ "-y", "@vernikr/[email protected]" ] } }}工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
這筆記錄未列出必要的環境變數。
@vernikr/browser Agent 透過 stdio 在本機執行。
上次檢查時間:2026-11-16。 MCP server that gives LLM agents a real headful Chrome (X11 mouse, Cloudflare Turnstile auto-solve) via the browser-agent-server daemon. Companion: github.com/vernikr/browser-agent