> ## Documentation Index
> Fetch the complete documentation index at: https://docs.laneapp.co/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Server

> Connect Lane to any MCP-compatible AI tool and let it read your workspace — Signals, Feedback, Plans, and Customers.

Lane's MCP server lets any [Model Context Protocol](https://modelcontextprotocol.io)-compatible assistant — Claude, [Cursor](/cursor), Windsurf, and others — read your Lane workspace directly. Ask about your product data from wherever you already work: "what are our top signals from enterprise customers?", "what's planned for Acme?", without copying anything into the prompt by hand.

Access is **read-only**: the server lets tools query and retrieve, never create or change anything in Lane.

## Availability

The MCP server is available on **Pro and Business** plans.

## Connect

1. Go to **Settings → MCP Server**.
2. Copy your **connection URL** — it includes a private token that authenticates your workspace.
3. Paste it into your assistant's MCP configuration as a remote server.

That's it — your tool can now query Lane. If you ever need to rotate access, use **Regenerate** to issue a new URL; the old one stops working immediately.

<Warning>
  Your connection URL contains a token that grants read access to your workspace. Treat it like a password — don't share it or commit it to a repo.
</Warning>

## What it can access

The server exposes eight read-only tools across the four areas an assistant needs to answer product questions. Each area has a **search** tool (find matching records) and a **get** tool (pull one record in full):

| Area          | Tools                              | What it returns                                                                                     |
| :------------ | :--------------------------------- | :-------------------------------------------------------------------------------------------------- |
| **Signals**   | `search_signals`, `get_signal`     | Recurring patterns, with priority, customers, and revenue                                           |
| **Feedback**  | `search_feedback`, `get_feedback`  | Individual feedback items and their Insights                                                        |
| **Plans**     | `search_plans`, `get_plan`         | Workspace and public Plans (private drafts aren't exposed)                                          |
| **Customers** | `search_customers`, `get_customer` | Account profiles — `get_customer` returns a full 360: profile, feedback, related signals, and plans |

Searches accept filters like a customer name, so an assistant can answer account-specific questions ("what does Acme want?") by scoping to that customer.

## FAQs

**Q: Can an assistant change my data through MCP?**\
A: No. Every tool is read-only — it can search and retrieve, but never create, edit, or delete.

**Q: Which tools does it expose?**\
A: Eight — a search and a get for each of Signals, Feedback, Plans, and Customers.

**Q: Are private Plans exposed?**\
A: No. Only workspace-visible and public Plans are returned; private drafts stay private.

**Q: How do I revoke access?**\
A: Regenerate the connection URL in **Settings → MCP Server**. The previous URL stops working right away.

**Q: What's the difference between this and the API?**\
A: MCP is for AI assistants to read your workspace conversationally. The [API](/api-key) is for programmatic integrations that create feedback and read signals.