Guide3 min read

Cursor + Ornold MCP: 3-Step Install via Agent Chat

Connect Cursor to Ornold MCP and automate antidetect browsers with AI. Step-by-step setup for Linken Sphere, Dolphin Anty, and more.

Apr 19, 2026

One Prompt — Cursor Installs Everything

You don't need to edit JSON, remember flags, or know which port your antidetect runs on. Paste one prompt with your token in Cursor Agent mode. Cursor reads the public install guide on GitHub, asks you 2 questions, then writes the right config to `~/.cursor/mcp.json`.

How it works: Ornold publishes a public agent guide at github.com/ornold-ai/ornold-mcp. Cursor fetches `AGENTS.md`, learns the install rules, and asks you only what it needs. No manual JSON, no copy-pasting between docs.

Step 1 — Get Your Token

Sign up at mcp.ornold.com
Open the Tokens page
Click Create token, copy the value (starts with `orn_sk_`)

Step 2 — Switch Cursor to Agent Mode

Open Cursor, open the chat panel, and check the dropdown at the top — switch from **Ask** to **Agent**. MCP tools and file edits only work in Agent mode.

Step 3 — Paste This Prompt

Paste this whole block into Cursor Agent chat:

Install Ornold MCP for me. I'm new to this, so guide me.

1. Fetch and read the official install guide:
   https://raw.githubusercontent.com/ornold-ai/ornold-mcp/main/AGENTS.md

2. Then ask me these questions in plain language:
   a) Which antidetect browser do I use? Options: Linken Sphere, Dolphin Anty, Vision Browser, Wadex, or none yet (just testing).
   b) Which interaction mode do I want? Briefly explain "vision" (AI sees the page like a human, recommended for antidetect) vs "dom" (AI reads HTML, faster but easier to detect). Recommend vision.
   c) If I picked Dolphin Anty or Vision Browser, ask me for the extra credentials (Dolphin API token / Vision token).

3. After I answer, edit ~/.cursor/mcp.json (create it if missing, merge with existing servers if present) using the Cursor JSON config format from AGENTS.md. Fill in my answers as flags in the args array. My Ornold token is: YOUR_TOKEN

4. Show me the final JSON before saving so I can confirm.

5. Tell me to restart Cursor, then to verify by asking "List my connected browsers using browser_list" in a new Agent chat.

If anything fails, explain the error in plain language.

Replace `YOUR_TOKEN` with your token. Cursor will fetch the guide, ask you 2-3 questions, edit `mcp.json`, and tell you to restart.

What Cursor Will Ask You

You don't need to memorize anything — but here's what to expect:

Question 1: Which antidetect browser?

**Linken Sphere** — desktop app, default port `40080`
**Dolphin Anty** — desktop app port `3001` + your API token from Dolphin dashboard
**Vision Browser** — cloud service, you provide their token (`vx_...`)
**Wadex** — desktop app, default port `8080`
**None** — install without antidetect, add later

Question 2: Vision or DOM mode?

**Vision (recommended)** — AI clicks via screenshots and coordinates. Antidetect-safe. Required for flow recording.
**DOM** — AI uses CSS selectors. Faster but anti-fraud detects this. Use only when stealth doesn't matter.
**Both** — two separate servers (advanced). Pick only if you really need both — too many tools confuse the agent.

Unsure? Pick **vision**. It's the safe default for antidetect.

Step 4 — Restart Cursor & Verify

After Cursor saves `mcp.json`, fully quit and reopen the app. Open a new chat in Agent mode and ask:

List my connected browsers using browser_list

If your antidetect is running, you'll see profiles. Done.

Already Know What You Want? Skip the Questions

Paste the exact JSON straight into `~/.cursor/mcp.json`. Examples:

Linken Sphere + vision (most common)

{
  "mcpServers": {
    "ornold-browser": {
      "command": "npx",
      "args": ["mcp-ornold-browser", "--token", "YOUR_TOKEN", "--linken-port", "40080", "--mode", "vision"]
    }
  }
}

Dolphin Anty + vision

{
  "mcpServers": {
    "ornold-browser": {
      "command": "npx",
      "args": ["mcp-ornold-browser", "--token", "YOUR_TOKEN", "--dolphin-port", "3001", "--dolphin-token", "YOUR_DOLPHIN_API_TOKEN", "--mode", "vision"]
    }
  }
}

Both modes (advanced)

{
  "mcpServers": {
    "ornold-vision": {
      "command": "npx",
      "args": ["mcp-ornold-browser", "--token", "YOUR_TOKEN", "--linken-port", "40080", "--mode", "vision"]
    },
    "ornold-dom": {
      "command": "npx",
      "args": ["mcp-ornold-browser", "--token", "YOUR_TOKEN", "--linken-port", "40080", "--mode", "dom"]
    }
  }
}

Public Resources

**GitHub** — github.com/ornold-ai/ornold-mcp — public agent install guide. Any AI agent can read it.
**npm** — mcp-ornold-browser
**Dashboard** — mcp.ornold.com

Troubleshooting

**MCP server not showing** — fully restart Cursor (quit + reopen, not just close window). Check `mcp.json` is valid JSON (no trailing commas).
**Browser tools missing in chat** — check the dropdown at top of chat — must be **Agent**, not **Ask**.
**Token invalid** — re-copy from mcp.ornold.com (full string, including `orn_sk_`).
**Cannot connect to browser** — antidetect must be running, port must match.
**npx: command not found** — install Node.js 18+ from nodejs.org, restart Cursor.

What's Next?

Parallel browser automation — run 10+ sessions at once
CAPTCHA solving guide — reCAPTCHA, hCaptcha, Turnstile
Vision AI vs CSS selectors — why coordinate clicks beat DOM scraping

Explainer