MCP Server

Point Claude, Cursor, or any MCP-compatible tool at your Orchly workspace and just ask. You get answers from the same numbers the dashboard shows, in whatever assistant you already work in.

Most of the time you open Orchly to look something up: a visibility score, who's pulling ahead, which sources AI keeps citing. The MCP server skips that trip. You ask your assistant in plain language, it reads your live Orchly data, and it answers, no dashboard, no exports.

Connect it in three steps

1. Add the server to your AI tool using the URL below. The setup guide has the exact steps per client.
2. Sign in once with your Orchly account and approve access. The session sticks around.
3. Start asking. Try "How has our AI visibility changed over the last 30 days?" and go from there.

Ask your data anything

A few things people reach for first:

• Track where you show up across ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Claude, and Grok.
• Size up competitors on visibility, sentiment, share of voice, and position, side by side.
• Find the sources AI trusts by pulling the domains and URLs it cites most, and spotting the ones that skip you.
• Read the why behind sentiment, the themes pushing your mentions positive or negative.
• Open up query fan-outs to see the sub-queries your prompts quietly trigger.
• Pull organic alongside AI: bot traffic, Search Console clicks, GA4, rank tracking, and page audit scores in one thread.
• Get the next move from planner recommendations, ranked.
• Change your setup by asking: add a prompt, spin up a topic, track a competitor, queue and run an agent task. It checks with you before anything sticks.

Want concrete examples? See use cases and prompts.

Where to point your client

Whatever tool you use, these three values are all it needs:

Sign in once

Pick whichever your client supports:

• OAuth 2.1. Your assistant pops open an Orchly consent page in the browser. Sign in, approve, done, it holds across conversations.
• API key. Mint a key in your Orchly settings and pass it as an Authorization: Bearer obk_live_xxx header. Handy for headless setups, CI, or clients without OAuth, and it's the same key the REST API takes.

Either route, the assistant only ever sees what your own workspace and permissions allow.

What it reads, what it changes

Think of it as mostly read-only: 26 of the 36 tools just fetch what's already there, visibility, citations, sentiment, traffic, and the rest.

The remaining 10 actually touch your setup, adding a prompt, creating a topic, tracking a competitor, kicking off a task. Read tools run on their own; write tools carry a hint that makes your client stop and ask before it commits. The tools reference lays out all 36.

Clients it works with

Anything that speaks the Model Context Protocol. We've written step-by-step setup for:

• Claude (Desktop, Web, Code)
• Cursor
• VS Code (GitHub Copilot)
• Windsurf

Grab the steps in the setup guide.

Before you connect

You'll need an Orchly account with at least one workspace, and a tool that supports MCP. That's it.