Documentation

Clerky MCP Server

Clerky works with AI assistants that support remote Model Context Protocol (MCP) servers, including Claude and ChatGPT. Once connected, the assistant can use available Clerky MCP tools with the team access you authorize.

The MCP server URL is:

https://mcp.clerky.com/mcp

Jump to the setup instructions for your AI tool:

Before You Begin

Before connecting, make sure you have:

  • A Clerky account you can sign in with.
  • Access to the Clerky teams you want the connector to use.

Keep in mind:

  • If you use Claude Team or Enterprise, or a ChatGPT Business, Enterprise, or Edu workspace, an owner or admin must add and approve the connector before members can install it.
  • Current Clerky MCP tools are read-only.
  • Team selection controls which Clerky teams the connector may access.
  • AI-generated answers can be incomplete or inaccurate. Review the underlying Clerky data before relying on assistant summaries for legal, tax, financing, or governance decisions.

Claude

Claude supports remote MCP servers as custom connectors.

Claude Web and Desktop

The steps are the same in Claude Web and Claude Desktop. On a personal account, you add the connector yourself:

  1. Open Settings > Connectors.

  2. Open the Add dropdown and choose Add custom connector.

  3. Enter a name: Clerky.

  4. Enter the connector URL:

    https://mcp.clerky.com/mcp

  5. Finish adding the connector.

  6. Click Connect, complete the Clerky OAuth flow, and choose the teams the connector may access.

If you use Claude Team or Enterprise, an organization owner or admin must add and approve the connector in Admin settings > Connectors before members can install it. Members don't add the connector themselves and won't see an Add custom connector option. Once the connector is approved, open Settings > Connectors, find Clerky, click Connect, complete the Clerky OAuth flow, and choose the teams the connector may access.

ChatGPT

ChatGPT supports apps and custom MCP apps.

If you use a ChatGPT Business, Enterprise, or Edu workspace, a workspace owner or admin must add and approve the connector before members can install it. In those workspaces, an owner or admin performs the setup steps below; after approval, members open Settings > Plugins, find Clerky, and connect their own Clerky account.

If Clerky is listed in the ChatGPT Plugins directory, open Settings > Plugins, find Clerky, and connect it.

To add Clerky as a custom MCP app, developer mode must be enabled first. The toggle is in Settings > Security and login; if it's unavailable, a workspace admin has to enable it. See OpenAI's developer mode guide for details.

ChatGPT Web

  1. In ChatGPT on the web, open Settings > Plugins > Browse Plugins (or go to chatgpt.com/plugins).

  2. Click the plus button to create a new plugin. The New Plugin dialog opens.

  3. Enter a name: Clerky, and optionally a short description.

  4. For Connection, keep Server URL selected and enter the MCP server URL: 

    https://mcp.clerky.com/mcp

  5. For Authentication, choose OAuth — Clerky MCP connections use OAuth only.

  6. Check I understand and want to continue to acknowledge OpenAI's warning about custom MCP servers. The Create button stays disabled until you do.

  7. Click Create.

  8. Complete the Clerky OAuth flow and choose the teams the app may access.

  9. Start a new chat, ask your question and name Clerky explicitly, for example: "Use Clerky to show the cap table for Acme, Inc."

ChatGPT may ask for confirmation before it uses apps, depending on your app permission settings.

ChatGPT Desktop

  1. In ChatGPT Desktop, open Settings > Plugins and switch to the MCPs tab.

  2. Click + Add Server.

  3. Enter a name: Clerky.

  4. Change the type to Streamable HTTP.

  5. Enter the MCP server URL:

    https://mcp.clerky.com/mcp

  6. Click Save.

  7. Complete the Clerky OAuth flow and choose the teams the app may access.

  8. Start a new chat, ask your question and name Clerky explicitly, for example: "Use Clerky to show the cap table for Acme, Inc."

Other AI Assistants

Cursor

Add the following configuration to the project's .cursor/mcp.json (or ~/.cursor/mcp.json to make Clerky available in all projects): 

{
  "mcpServers": {
    "clerky": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote@latest",
        "https://mcp.clerky.com/mcp"
      ]
    }
  }
}

Other MCP clients

If your assistant supports remote MCP servers, you can use the Streamable HTTP transport: 

https://mcp.clerky.com/mcp

On first use, the MCP client should open a browser-based OAuth flow so you can sign in and authorize team access.

What You Can Ask

Current Clerky MCP tools help you identify an authorized team and retrieve cap table and corporate representation data for that team.

For example, you can ask:

  • Check cap table ownership: "Who owns what on our cap table, and what's my ownership percentage?"
  • Review fundraising history: "How much have we raised through SAFEs, and what were the key terms?"
  • Track available equity: "How much equity remains in our option pool?"
  • Model exit proceeds: "If we sold the company for $1B, roughly how much would each stockholder receive, assuming all SAFEs and notes convert?"
  • Review corporate representation: "Show the corporate representation data for this company."

If your assistant is also connected to your accounting or banking tools, you can combine them with Clerky data:

  • Reconcile equity compensation (with a Xero or QuickBooks connector): "Cross-check our equity grants against our financial statements in Xero."
  • Verify investor funding (with a Mercury, Brex, or Rho connector): "Compare our SAFE issuances against incoming wires in Mercury."

The Clerky MCP server currently provides:

Tool Description
list_teams Lists the Clerky teams available through the current authorization.
get_team_cap_table Returns cap table and corporate representation data for the selected team.

Because these results may include sensitive company, equity, and representation information, use get_team_cap_table only when that data is needed for the current task.

Security

Treat cap table and corporate representation data as confidential.

  • OAuth only. Clerky MCP uses OAuth. You don't paste Clerky credentials or access tokens into prompts — the MCP client handles OAuth and token storage. When an assistant calls a Clerky tool, Clerky authorizes the request using your OAuth authorization and the selected teams.
  • Scope access to the teams you need. The connector can use Clerky tools only with the teams you authorize. You can deny the request instead of authorizing it.
  • Mind the assistant provider's terms. Tool results may be included in the assistant conversation and handled according to the provider's terms, data controls, and workspace settings. Use provider settings appropriate for your organization.
  • Verify before you rely on it. Review Clerky source data before relying on assistant-generated summaries or analysis.

Authentication

Clerky MCP server connections use OAuth. The MCP client opens a browser-based Clerky login and authorization flow the first time you connect.

During authorization, Clerky asks which teams to include. You can select individual teams or allow access to all teams. Select only the teams you want the connector to use.

Troubleshooting

Match the symptom below.

  • I don't see Clerky tools. Enable Clerky for the current conversation, then start a new chat. Some assistants load MCP tools only when a conversation starts.
  • Authentication doesn't complete. Disconnect and reconnect Clerky in the assistant's settings. Complete OAuth in the same browser session where you are logged in to Clerky.
  • The assistant picked the wrong team. Ask the assistant to call list_teams, then tell it exactly which team to use before asking for cap table or corporate representation data.
Previous
Authentication