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

# Connect an AI assistant

> Connect Claude or Codex to review Candidates and interview results from your AI assistant.

MCP (Model Context Protocol) connects InterviewFlowAI to an AI assistant such as Claude or Codex. You can list AI Interviewers, review Candidate details, summarize interviews, update Candidate notes, and change supported Interviewer settings without switching applications.

<Warning>
  Sign in to MCP with the same email address you use for your InterviewFlowAI account. This connects the authenticated MCP session to the correct Company. If the connector asks you to create an account, use the same email address and verify it. Check your spam or junk folder for the verification email.
</Warning>

## Before you start

* Your Company needs an active subscription or trial. If it has expired, the MCP server refuses the connection.
* Each connection can make up to 15 requests per minute. If your assistant goes over the limit, wait a minute and try again.

The InterviewFlowAI MCP server uses this URL:

```text theme={null}
https://api.interviewflowai.com/mcp
```

## Choose your MCP client

Follow the setup for your client, then sign in with the same email address as your InterviewFlowAI account.

<Tabs>
  <Tab title="Claude Code CLI">
    <Steps>
      <Step title="Add the InterviewFlowAI server">
        Run this command in your terminal:

        ```bash theme={null}
        claude mcp add-json interviewflowai '{"type":"http","url":"https://api.interviewflowai.com/mcp","oauth":{"clientId":"ACYUUGM2Ia9tbufCbwQSELKlyTvlgxem","callbackPort":8765,"scopes":"mcp:read mcp:write"}}'
        ```
      </Step>

      <Step title="Authenticate">
        Start Claude Code, run `/mcp`, select **interviewflowai**, and choose **Authenticate**.
      </Step>

      <Step title="Confirm the connection">
        Run `/mcp` again and confirm that the server shows as connected.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Claude Desktop">
    <Steps>
      <Step title="Add a custom connector">
        Open **Customize → Connectors**, click **+**, and select **Add custom connector**. Enter **InterviewFlowAI** as the name and `https://api.interviewflowai.com/mcp` as the server URL, then click **Continue**.

        ![Claude Add custom connector form with the InterviewFlowAI name and MCP URL](https://website-images.interviewflowai.com/docs/claude-desktop-add-connector.png)
      </Step>

      <Step title="Configure authentication">
        Select **Always required** under **Authentication**. Select **Use your own OAuth client** under **OAuth client**.

        ![Claude connector authentication settings with Always required and Use your own OAuth client selected](https://website-images.interviewflowai.com/docs/claude-desktop-oauth-options.png)
      </Step>

      <Step title="Enter the OAuth client ID">
        Enter `ACYUUGM2Ia9tbufCbwQSELKlyTvlgxem` as the Client ID. Leave **OAuth Client Secret** blank, then click **Add**.

        ![Claude connector form showing the InterviewFlowAI OAuth Client ID and an empty client secret](https://website-images.interviewflowai.com/docs/claude-desktop-client-id.png)
      </Step>

      <Step title="Authenticate">
        Select **Connect** or **Authenticate** for InterviewFlowAI and complete the sign-in. Enable InterviewFlowAI from the **Connectors** menu in your conversation.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Codex CLI">
    <Steps>
      <Step title="Configure the MCP server">
        Open or create the Codex configuration file for your operating system:

        * **macOS and Linux:** `~/.codex/config.toml`
        * **Windows PowerShell:** `$HOME\.codex\config.toml`
        * **Windows Command Prompt:** `%USERPROFILE%\.codex\config.toml`

        Add this configuration:

        ```toml theme={null}
        [mcp_servers.interviewflowai]
        url = "https://api.interviewflowai.com/mcp"
        scopes = ["mcp:read", "mcp:write"]

        [mcp_servers.interviewflowai.oauth]
        client_id = "ACYUUGM2Ia9tbufCbwQSELKlyTvlgxem"
        callback_port = 52000
        callback_url = "http://127.0.0.1:52000/callback"
        ```
      </Step>

      <Step title="Authenticate">
        Run this command, then complete the sign-in in your browser:

        ```bash theme={null}
        codex mcp login interviewflowai
        ```
      </Step>

      <Step title="Verify and restart">
        Run this command and confirm that the server is enabled:

        ```bash theme={null}
        codex mcp get interviewflowai
        ```

        Restart Codex so it loads the InterviewFlowAI tools.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Codex Desktop">
    <Steps>
      <Step title="Configure the MCP server">
        Quit Codex Desktop, then open your Codex configuration file:

        * **macOS and Linux:** `~/.codex/config.toml`
        * **Windows PowerShell:** `$HOME\.codex\config.toml`

        Add this configuration:

        ```toml theme={null}
        [mcp_servers.interviewflowai]
        url = "https://api.interviewflowai.com/mcp"
        scopes = ["mcp:read", "mcp:write"]

        [mcp_servers.interviewflowai.oauth]
        client_id = "ACYUUGM2Ia9tbufCbwQSELKlyTvlgxem"
        callback_port = 52000
        callback_url = "http://127.0.0.1:52000/callback"
        ```
      </Step>

      <Step title="Authenticate">
        Open Codex Desktop, go to **Settings → MCP Servers**, and confirm that **interviewflowai** appears as a Streamable HTTP server. Select **Authenticate** and complete the sign-in.
      </Step>

      <Step title="Restart and verify">
        Restart Codex Desktop. Type `/mcp` in the composer and confirm that InterviewFlowAI is connected.
      </Step>
    </Steps>
  </Tab>
</Tabs>

## Other assistants

For any other assistant or an API key connection, ask your technical team or [support](mailto:support@interviewflowai.com) to help you connect.

<Warning>
  An API key connection has full **Owner** access, including write actions, even if you normally use InterviewFlowAI as a **Viewer**. Keep the key private and delete it if it is exposed.
</Warning>

## Permissions

Your assistant can list and inspect Interviewers and Candidates. **Owners** can also update Candidate notes, custom fields, Interviewer visibility, and archived status. **Viewers** have read-only access.

Confirm every write action before you allow your assistant to make a change.

## Try the connection

Ask your assistant:

* “Show me the Candidates who completed the interview.”
* “Show me the top 10 Candidates by interview score.”
* “Show me the Candidates with an interview score of 70 or above.”


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.