Note: Greenhouse MCP is currently in open beta. Access depends on your subscription tier and admin configuration.
Site Admins can connect Greenhouse MCP to Claude, Claude Code, ChatGPT, or Gemini CLI using the steps in this article. Greenhouse MCP also supports Amazon Q, Grok AI, and other compatible AI tools that support OAuth 2.0 with PKCE and Dynamic Client Registration. For an overview of what Greenhouse MCP is and how it works, read the main Greenhouse MCP article.
Before you begin
Confirm the following before starting setup:
- Your organization is on a Core, Plus, or Pro subscription tier.
- You are a Site Admin in Greenhouse Recruiting.
- Your AI tool and Greenhouse Recruiting account must use the same email address.
- Gemini supports MCP connections through Gemini CLI only. The Gemini web app does not support remote MCP server connections.
Set up Greenhouse MCP with Claude
To add a custom connector, you may need your organization's Claude account owner to grant you permission or complete setup on your behalf.
- Open claude.ai and sign in.
- In the left menu bar, click your profile at the bottom and select Settings.
- Click Connectors, then click Add custom connector.
- Enter a name (for example, Greenhouse MCP) and a description.
- In the Remote MCP server URL field, enter:
https://mcp.us.greenhouse.io/mcp - Click Add.
- On the Connectors page, click Connect next to the Greenhouse MCP entry.
- Sign in to Greenhouse when prompted, review the requested permissions, and click Authorize.
You'll be redirected back to Claude. Greenhouse Recruiting is now connected, and Claude can access Greenhouse tools directly in your conversations.
Tip: To connect to a Greenhouse sandbox instead of your production account, sign out of Greenhouse and close the app before following the steps above.
Set up Greenhouse MCP with Claude Code
Claude Code users can add the Greenhouse MCP server from the terminal.
Step 1: Add the MCP server
Open your terminal and run:
claude mcp add greenhouse --transport http https://mcp.us.greenhouse.io/mcp
Step 2: Authenticate
A browser window may open prompting you to sign in and authorize access. Complete the authentication flow, then return to your terminal.
Step 3: Verify the connection
Inside a Claude Code session, run:
/mcp
You should see greenhouse listed as a connected server. If it shows as connected, setup is complete.
Set up Greenhouse MCP with ChatGPT
Greenhouse MCP connects to ChatGPT through the Apps settings. Apps can only be created when Developer Mode is enabled in your ChatGPT account.
- Open chatgpt.com and sign in.
- In the left menu bar, click your profile at the bottom and select Apps.
- Click Create.
- Enter a name (for example, Greenhouse MCP) and a description.
- In the MCP Server URL field, enter:
https://mcp.us.greenhouse.io/mcp - Leave Authentication set to OAuth — do not change this setting.
- Agree to the terms and click Create.
- You'll be redirected to Greenhouse. Sign in when prompted, review the requested permissions, and click Authorize.
You'll be redirected back to ChatGPT, where Greenhouse Recruiting should appear as Enabled with a green confirmation banner.
Tip: To connect to a Greenhouse sandbox instead of your production account, sign out of Greenhouse and close the app before following the steps above.
Set up Greenhouse MCP with Gemini CLI
Greenhouse MCP connects to Gemini through Gemini CLI only. The Gemini web app does not support remote MCP server connections.
Step 1: Add the MCP server
Open your terminal and run:
gemini mcp add greenhouse --transport http https://mcp.us.greenhouse.io/mcp
Step 2: Authenticate
A browser window may open prompting you to sign in and authorize access. Complete the authentication flow, then return to your terminal.
Step 3: Verify the connection
Inside a Gemini CLI session, run:
/mcp auth greenhouse
You should see greenhouse listed as a connected server. If it shows as connected, setup is complete.
Tip: To connect to a Greenhouse sandbox instead of your production account, sign out of Greenhouse and close the app before following the steps above.
Manage scopes in Dev Center > MCP Access
Site Admins control which Greenhouse API scopes are available to users in their organization. Scopes determine which categories of Greenhouse data and actions — such as reading candidates, managing jobs, or editing applications — are accessible through the MCP. Users cannot grant themselves access beyond what the admin has enabled.
- Navigate to Dev Center > MCP Access.
- Review the list of available scopes.
- Enable the scopes you want to make available to users.
- Click Save.
Scopes that are not enabled will not appear when users authorize their AI tool, and the corresponding tools will not be available. Return to this page at any time to update the configuration.
Note: If you change the enabled scopes after users have already connected their AI tool, those users will need to disconnect and reconnect to apply the changes.
Troubleshooting
Wrong Greenhouse account was used during authorization
Disconnect the MCP integration from your AI tool and reconnect. Sign in with the correct Greenhouse account when prompted to authorize.
Connected to sandbox instead of production (or vice versa)
Greenhouse MCP connects to whichever account you're signed in to at the time of authorization. To switch, disconnect the integration, sign out of Greenhouse, then reconnect and sign in to the correct account.
Expected tools or capabilities aren't appearing
Check that the relevant scopes are enabled in Dev Center > MCP Access. Tools that correspond to disabled scopes will not appear for users.
Capabilities changed after the admin updated scopes
If an admin enables or disables scopes after you've already connected, disconnect and reconnect your AI tool to apply the updated permissions.
Gemini web app isn't working
The Gemini web app does not support remote MCP server connections. Complete setup through Gemini CLI instead.
Frequently asked questions
Who can connect Greenhouse MCP to an AI tool?
Only Site Admins can complete the initial setup.
Can users grant themselves access to scopes the admin hasn't enabled?
No. The scopes enabled in Dev Center > MCP Access act as a ceiling. Users can only authorize access within what the admin has already enabled.
Do users need to reconnect after the admin changes scopes?
Yes. Users who are already connected will need to disconnect and reconnect to pick up any scope changes.
Is the Gemini web app supported?
No. Greenhouse MCP connects to Gemini through Gemini CLI only.
What can I do once Greenhouse MCP is connected?
With Greenhouse MCP connected, you can use your AI tool to access Greenhouse data through natural language — for example, listing open jobs, reviewing candidates, or checking application stages. For a full overview of capabilities, read the main Greenhouse MCP article and learn about common MCP use cases by role.