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

# Codex

> Connect the Glasser MCP server to local Codex clients.

Local Codex clients share MCP settings on the same host. You can use the connection in the
app, CLI, or IDE extension. For a Skill with CLI instructions, use the [Codex Skill guide](/docs/setup-skill/codex).

These steps use the Codex CLI to configure the connection for your local clients.

<Steps>
  <Step title="Add the server">
    Run this in a terminal where the Codex CLI is installed:

    ```bash theme={null}
    codex mcp add glasser --url https://api.glasser.ai/mcp
    ```

    Codex saves this server in `~/.codex/config.toml` by default.
  </Step>

  <Step title="Sign in">
    Follow the OAuth prompt if it appears during setup. Otherwise, run:

    ```bash theme={null}
    codex mcp login glasser
    ```

    In the browser, sign in to Glasser, choose your Workspace, and select **Allow**.
    Create an account and Workspace if needed.
  </Step>

  <Step title="Start a new local task">
    Reopen the client if needed so it loads the updated configuration. In the Codex CLI,
    `/mcp` shows the active servers; check that `glasser` is available.
  </Step>

  <Step title="Check the connection">
    In a new conversation, ask:

    ```text wrap theme={null}
    Use Glasser to check my available balance. Do not make any paid Runs.
    ```

    The agent should use Glasser's `balance` tool and return your balance. This check is free;
    a zero balance still means the connection works. If you need funds for a paid Run, add
    them in the [Glasser console](https://app.glasser.ai).
  </Step>
</Steps>

## Try your first request

```text wrap theme={null}
Use Glasser to find recent news about Stripe. Show me the data you can retrieve and the price, then wait for my approval before any paid Run.
```

After you approve, the agent can retrieve the data and report the result and charge.

See [OpenAI's MCP documentation](https://learn.chatgpt.com/docs/extend/mcp?surface=cli) for
client settings and authentication help.


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