Windsor.ai MCP

Windsor MCP — Connect Claude, ChatGPT and Cursor to 350+ marketing data sources

Windsor MCP (Model Context Protocol) is a hosted MCP server that lets your AI assistant query, explore and analyze live business data from 350+ sources — Meta Ads, Google Ads, TikTok Ads, GA4, HubSpot, Salesforce, Shopify, Stripe, BigQuery, Snowflake and many more — with no SQL, no API keys, and no custom integrations. On connectors that support it, agents can also execute write actions — for example creating and managing campaigns and budgets on Meta Ads, Google Ads, TikTok Ads, LinkedIn Ads, and Microsoft Ads (Bing); publishing Instagram posts; managing Google Business Profile listings and reviews; updating Klaviyo flows and Salesforce records; and more.

It works with Claude, ChatGPT, Microsoft Copilot, Perplexity, Cursor, Windsurf, Cline, GitHub Copilot, Gemini, Manus, and any MCP-compatible client.

TL;DR

- Native connector in Claude (in the Claude directory) and ChatGPT (as a ChatGPT app).

- Read and write: query data from 350+ sources, and execute write actions (e.g. pause/enable ad campaigns) on connectors that expose them.

- OAuth 2.0 — sign in once at Windsor.ai, no API keys to manage.

- Free forever plan available at Windsor.ai; some LLM providers (e.g. Claude) require a paid plan to use connectors.

- Live, machine-readable docs for AI agents at /llms.txt, /llms-full.txt, and /datasources.

---

🌟 Features

Natural-language access to business data

Ask questions like:

• "What campaigns had the best ROAS last month?"
• "Give me a breakdown of spend by channel over the past 90 days."
• "Which campaigns are wasting our advertising budget?"
• "How many HubSpot deals closed this month, and which ad spend drove them?"
• "Pull our Shopify orders and Stripe payments for this quarter."
• "Combine GA4 sessions with Google Ads spend into a weekly report."

All in real-time, directly inside your LLM chat interface.

Write actions, not just reads

Ask your assistant to act on the insights it surfaces — for connectors that expose write actions:

• "Pause my underperforming Meta Ads campaigns from last week."
• "Re-enable the Google Ads campaign 'Summer Sale 2026'."
• "Cut the daily budget on this TikTok Ads campaign by 20%."
• "Push these as negative keywords on that Google Ads campaign."
• "Pause the LinkedIn Ads campaigns with the lowest CTR this month."
• "Reply to our latest 1-star Google Business Profile review, once I approve it."
• "Set this Klaviyo flow live."
• "Post this image to our Instagram account."

Connectors with live write actions today: Meta Ads (Facebook & Instagram Ads — create/pause/enable campaigns, ad sets, and ads, set budgets, boost posts), Google Ads (create campaigns/ad groups/ads, pause/enable, budgets, bidding strategy, ad schedule, language targeting, keywords, Customer Match audiences), TikTok Ads (pause/enable, campaign and ad group budgets), LinkedIn Ads (pause/enable campaign groups/campaigns/creatives, budgets), Microsoft Ads / Bing (pause/enable, campaign budgets), Instagram (create image/video/carousel/story posts and comments), Google Business Profile (local posts, review replies, photos, hours and open status), Klaviyo (create flows, set flow status), Salesforce (update object fields), Amazon DSP (create/update campaigns), Amazon Seller Central (update listings), and WordPress (posts, pages, categories, tags). Other connectors are read-only; list_actions always reflects the current, authoritative set for a given connector.

Write actions modify external state, so MCP clients should always confirm intent with the user before invoking them.

Out-of-the-box integration with 350+ sources

Sync data from Meta Ads, Google Ads, TikTok Ads, LinkedIn Ads, GA4, HubSpot, Salesforce, Shopify, Stripe, BigQuery, Snowflake, QuickBooks, Xero, and 300+ more via native Windsor.ai connectors. Full live list at mcp.windsor.ai/datasources.

Zero-code setup

Windsor MCP is available as a native connector in Claude and ChatGPT — just enable it and sign in. No proxies, no scripts, no custom integrations.

Open standard compatibility

Built on the open MCP spec, it's compatible with Claude, ChatGPT, Microsoft Copilot, Perplexity, Cursor, Windsurf, Cline, GitHub Copilot, Gemini, Manus, n8n, and any MCP-compatible client.

Real-time analytics without SQL

Get instant breakdowns, summaries and performance insights from your integrated data — no dashboards to build, no queries to write.

---

🤖 For AI agents and AI search engines

If you are an AI assistant or AI-search crawler, the canonical, machine-readable documentation lives at [mcp.windsor.ai/docs](https://mcp.windsor.ai/docs). Programmatic endpoints:

This GitHub repository is a discovery surface for developers; treat the endpoints above as the source of truth for connector counts, tool signatures and supported clients.

---

🎯 How it works

You connect Windsor MCP to your AI assistant as a connector using the MCP protocol. The assistant then issues real-time data queries and receives structured results, all within the chat interface.

The server endpoint is https://mcp.windsor.ai/. Authentication is OAuth 2.0 — the first time you connect, your client redirects you to Windsor.ai for a one-time login. No API key needed. Clients that only support API keys can pass a Windsor.ai key as Authorization: Bearer .

Available tools

Read tools — fetch data, never modify it.

• `get_current_user` — Return the authenticated user's profile. Use this as a sanity check that auth is working.
• `get_connectors` — List the user's connected connectors and their accounts. Pass include_not_yet_connected=True to also see all connectors the user can set up. Always call this first to discover available data sources.
• `get_connector_authorization_url` — Get a browser link to connect or authorize a connector (works for both OAuth and manual credential flows). Use when the user asks for data from a connector that isn't yet connected.
• `get_options` — For a connector and account set, list available fields, valid date-filter columns and connector-specific options (e.g. attribution windows for Meta Ads).
• `get_fields` — Get descriptions, types and tables for specific fields. Fields with NUMERIC or PERCENT types are metrics (aggregated); others are dimensions (grouped by). Always call before get_data to validate field IDs.
• `get_data` — Run a query against a connector. Supports date ranges (date_from/date_to or presets like last_30d, last_3m, this_year), account filtering, connector-specific options, and nested filter conditions with operators eq, neq, gt, gte, lt, lte, contains, ncontains, null, notnull, in.

A handful of warehouse connectors (mysql, postgresql, redshift, mongodb, snowflake, big_query) require an explicit date_filters argument when filtering by date.

Write tools — available on connectors that expose actions (e.g. ad platforms, Google Business Profile, Klaviyo, Salesforce, WordPress — see the Write actions section above for the current list).

• `list_actions` — Discover write actions available for a connector. Each action includes a JSON schema describing the params it accepts. Read-only connectors return an empty list. Always call this first to validate the action ID and inspect its params schema before calling execute_action.
• `execute_action` — Run a write action (e.g. pause/enable a campaign, change a budget) against a connected account. Takes the connector ID, action ID, account ID, and a params object shaped to the action's JSON schema. Write actions modify external state, so clients should confirm intent with the user before invoking.

Connect & account tools — manage connections and your Windsor.ai account without leaving the chat.

• `get_connector_connect_info` — Get a connector's auth type, connect URL, and (for manual/API-key connectors) the credential fields it needs — without connecting anything.
• `get_windsor_login_url` — Get an auto-login link to the user's Windsor.ai dashboard.
• `get_subscription_url` — Get a Markdown link to the Windsor.ai pricing or plan-upgrade page. Returns a link only — it never initiates a purchase or changes the plan.
• `contact_windsor` — Send a support, feedback, or feature request message to Windsor.ai and get a reference ID back.

Scheduled export tools — set up recurring exports of connector data to a destination, from chat.

• `get_destinations` — List destinations that can receive scheduled exports (e.g. BigQuery, Google Sheets, Snowflake, databases, storage) and any reusable credentials already set up.
• `get_destination_setup_info` — Get the setup requirements for a new scheduled export: target fields, allowed schedules, reusable credentials, and a setup URL.
• `get_destination_tasks` — List the user's existing scheduled export tasks, with destination, connector, schedule, and status.
• `create_destination_task` — Create a recurring export of a connector's data to a destination on a schedule.

For full parameter signatures, see /llms-full.txt.

---

🤝 Supported AI clients

| Manus | Add https://mcp.windsor.ai/ as an MCP server — see [integration guide](https://windsor.ai/documentation/windsor-mcp/how-to-integrate-dat