Quickstart

Get Adros running in your AI tool in under 3 minutes. Same Adros brain, any client.

What you need first

1. An Adros account. Sign up at app.adros.ai (opens in a new tab). Free tier gives you a working MCP token for testing.
2. Your MCP token. From the dashboard → Settings → MCP → copy the lyr_... token. Treat it like a password — anyone with this token can call Adros as you.
3. At least one connected ad account (optional for now, required for any real work). Connect Meta and/or Google Ads from the dashboard once you're in.

MCP endpoint: https://api.adros.ai/mcp

Claude Desktop

The fastest path — Claude Desktop natively supports MCP over Streamable HTTP.

1. Open your Claude Desktop config:

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

2. Add Adros to the mcpServers section:

{
  "mcpServers": {
    "adros": {
      "url": "https://api.adros.ai/mcp",
      "headers": {
        "X-User-Token": "lyr_YOUR_TOKEN_HERE"
      }
    }
  }
}

3. Restart Claude Desktop. You should see a hammer icon at the bottom of the chat input — click it to verify Adros tools are loaded. You should see 74 tools.

4. Test it. In a new chat, say:

"Use Adros to list my clients and tell me which ones are in 'optimizing' state."

Claude will call list_clients + get_client_state automatically.

Cursor

Cursor added native MCP support in late 2025.

1. Open Cursor Settings → Features → MCP.

2. Click "Add new MCP server":

• Name: adros
• Type: http (Streamable HTTP, not stdio)
• URL: https://api.adros.ai/mcp
• Headers: X-User-Token: lyr_YOUR_TOKEN_HERE

3. Save and restart Cursor. The tools will be available in Composer and Chat.

4. Test in Composer:

"Audit my Meta ads for creative fatigue using Adros."

OpenClaw / Paperclip

If you're running OpenClaw as your agent orchestration layer, add Adros as an MCP server in your ~/.openclaw/config.yaml (or equivalent):

mcp_servers:
  adros:
    transport: streamable_http
    url: https://api.adros.ai/mcp
    headers:
      X-User-Token: lyr_YOUR_TOKEN_HERE

Also set an environment variable so your Python workers can hit the REST API directly:

# ~/.openclaw/.env
ADROS_BASE_URL=https://api.adros.ai/api/v1
ADROS_API_KEY=lyr_YOUR_TOKEN_HERE

⚠️ Important: the REST base URL MUST include /api/v1 at the end. Without it, every call returns 404. See Troubleshooting for the most common setup mistake.

Any MCP-compatible client

Adros speaks Streamable HTTP MCP (spec version 2025-03-26). Any MCP client that supports this transport will work.

Connection parameters:

Raw curl test (no client needed)

If you want to verify your token works before setting up a client, hit a REST endpoint directly:

curl -H "Authorization: Bearer lyr_YOUR_TOKEN_HERE" \
  "https://api.adros.ai/api/v1/contract/specs?stage=optimizing"

Expected response: JSON with "schema_version":"v1" and "total":4.

If you get a 401 — token is wrong. If you get a 404 — you probably dropped the /api/v1 prefix. If you get a 200 with real data — you're connected.

Next steps

• Read the Technical Overview to understand what Adros actually does (4,022 patterns, 74 tools, daily monitor, contract layer).
• Browse the Tool Catalog to see every MCP tool grouped by category.
• Skim the FAQ for the 5 most common setup mistakes.
• Go to app.adros.ai (opens in a new tab) and connect your Meta + Google Ads accounts so Adros has data to work with.

Troubleshooting the first connection

See the full FAQ / Troubleshooting page for deeper fixes.