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

# Scispot MCP Server

> Connect an AI assistant to your Scispot workspace with the Scispot MCP Server.

The Scispot MCP Server connects AI assistants to your Scispot workspace. You can ask questions, retrieve lab data, and perform supported actions in natural language.

<Warning>
  The Scispot MCP Server is in private beta. Contact your Customer Success representative to request access before you connect a client.
</Warning>

## Choose MCP or the REST API

Use the MCP Server when you want an AI assistant to work with Scispot conversationally. Use the [REST API](/api-reference/overview) when you build a programmatic integration or automation.

## Before you begin

You need:

* Scispot MCP beta access
* A Scispot account with access to the data you want to use
* A [Personal Token](/authentication) for clients that use Bearer authentication
* An MCP-compatible AI client

The MCP Server uses `https://mcp.scispot.io`.

## Security and permissions

The MCP Server respects your existing Scispot permissions. You can access only the data and actions that your Scispot account allows.

Web connectors use OAuth. Developer clients can authenticate with your Personal Token in the `Authorization: Bearer` header.

<Warning>
  Treat your Personal Token like a password. Do not paste it into prompts, source control, screenshots, or shared configuration files. External AI providers handle any data sent to them under their own terms and your organization's data-handling policies.
</Warning>

## Connect Claude.ai

Use Claude.ai's custom connector flow to connect through OAuth.

<Steps>
  <Step title="Open custom connectors">
    In Claude.ai, open **Settings** and then **Connectors**. Select **Add custom connector**.
  </Step>

  <Step title="Enter the server URL">
    Enter `https://mcp.scispot.io` as the remote MCP server URL.
  </Step>

  <Step title="Configure OAuth">
    Open the connector's advanced OAuth settings. Enter `claude-ai-client` as the client ID. Generate a Scispot Personal Token and enter it as the client secret.
  </Step>

  <Step title="Authorize Scispot">
    Complete the OAuth flow and approve the connection. Claude can then discover the Scispot tools that your account can use.
  </Step>
</Steps>

Test the connection with a read-only request such as: “List my Scispot Labsheets.”

## Connect Claude Code

Claude Code supports remote MCP servers over HTTP. Use the interactive MCP setup flow and configure the Scispot server as an HTTP MCP server.

<Steps>
  <Step title="Start MCP setup">
    Run `claude mcp add` in your terminal and follow the interactive prompts.
  </Step>

  <Step title="Choose an HTTP server">
    Select an HTTP or Streamable HTTP MCP server when Claude Code asks for the transport.
  </Step>

  <Step title="Enter the server URL">
    Enter `https://mcp.scispot.io` as the server URL.
  </Step>

  <Step title="Add your authorization header">
    Add an `Authorization` header with the value `Bearer &lt;your-personal-token&gt;`.
  </Step>

  <Step title="Verify the connection">
    Start or restart Claude Code. Ask it to list your Scispot Labsheets to confirm that it can discover the server tools.
  </Step>
</Steps>

## Connect another MCP-compatible client

You can connect other clients that support remote HTTP MCP servers. The exact labels and configuration experience vary by client.

1. Add a remote MCP server or connector in your client.
2. Set the server URL to `https://mcp.scispot.io`.
3. Choose OAuth when your client supports the Scispot OAuth flow.
4. Otherwise, configure an `Authorization` header with `Bearer <your-personal-token>`.
5. Confirm the connection with a read-only request before you perform a write operation.

Contact Customer Success if your client cannot complete the OAuth flow or does not support custom HTTP headers.

## What you can do

The MCP Server provides 33 tools across these areas:

| Area           | Examples                                                                |
| -------------- | ----------------------------------------------------------------------- |
| **Labsheets**  | List Labsheets, inspect schemas, search rows, and add or update rows    |
| **ELN**        | Find, create, and update experiments, protocols, and documentation      |
| **Inventory**  | Retrieve manifests and browse locations                                 |
| **Labflows**   | List samples in a Labflow                                               |
| **Media**      | Retrieve thumbnails, upload images, and attach images to Labspace pages |
| **Smart Apps** | List, create, and publish supported Smart Apps                          |

The tools that change data require the same permissions as the equivalent action in Scispot.

## Use write actions safely

Start with read-only requests. Review the data and planned action before you approve a write request.

For important records, ask the AI to summarize the proposed changes before it creates or updates data. Review the resulting record in Scispot after the action completes.

For regulated workflows, make sure your use of the AI client and its data handling meets your organization's validation, review, and approval requirements.

## Troubleshooting

### You cannot add the connector

Confirm that Customer Success enabled MCP beta access for your workspace. Verify that you entered `https://mcp.scispot.io` exactly.

### Authentication fails

Generate a new [Personal Token](/authentication) and confirm that the header uses the format `Authorization: Bearer <your-personal-token>`. Do not include quotation marks around the token.

### You cannot see data or perform an action

The MCP Server uses your Scispot permissions. Confirm that you can complete the same action in Scispot with your account.

### The AI proposes an unexpected change

Do not approve the action. Ask the AI to explain the proposed change, then verify the target records in Scispot before you continue.