Skip to main content

Agent setup

Connect your AI coding agent to Market Data

Sign up for the free trial, then paste one prompt into the tool you already use. The agent reads our llms.txt and does the rest: the Sheets add-on or the SDK for your language, your token from the environment, and a first live call.

The prompt works in

  • Claude Code
  • Codex
  • Cursor
  • GitHub Copilot
  • Gemini CLI
  • Claude Desktop
  • ChatGPT

Paste into your coding agent

Read https://www.marketdata.app/llms.txt and set up Market Data. Decide if I want data in a spreadsheet or in code; if unclear, ask one question. Spreadsheet: set up the Google Sheets add-on and its formulas (STOCKDATA, OPTIONDATA, OPTIONCHAIN, EARNINGS) and link its docs. Code: start with the official Market Data SDK for my language (Python, JavaScript, Java, C#, Go, PHP; https://www.marketdata.app/docs/sdk/); never write your own HTTP client while an SDK exists. If I have no account, tell me to sign up for the free trial at https://www.marketdata.app/signup/ and set MARKETDATA_TOKEN myself; you cannot do that step. Read the token from MARKETDATA_TOKEN, never paste it in chat or source, then make one real call (an SDK stock quote or one formula in a cell) to confirm it works.

Open in Claude (opens in a new tab) Open in ChatGPT (opens in a new tab)

Prefer to do it yourself? Start the free trial or read the API docs.

Pick your agent

Choose the tool you work in. Each card opens the steps for that tool, and the first step is always yours: the agent cannot sign up or fetch a token for you.

  • Terminal agent

    Claude Code

    Anthropic's coding agent, run from a terminal inside your project.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open a terminal in the project directory and start Claude Code.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Terminal agent

    Codex

    OpenAI's coding agent, run from a terminal inside your project.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open a terminal in the project directory and start Codex.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Editor agent

    Cursor

    A code editor with an agent built in, working on the files you have open.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open the project in Cursor and open its chat panel.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Editor agent

    GitHub Copilot

    GitHub's coding assistant, running inside the editor you already use.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open the project in your editor with Copilot enabled and open the Copilot chat.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Terminal agent

    Gemini CLI

    Google's coding agent, run from a terminal inside your project.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open a terminal in the project directory and start Gemini CLI.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Chat assistant

    Claude Desktop

    Anthropic's desktop app. It writes the setup for you to add to your project or sheet.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Start a new conversation and paste the prompt, or use the Open in Claude link above to start one in the browser.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Chat assistant

    ChatGPT

    OpenAI's assistant in the browser or the desktop app. It writes the setup for you to add.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Start a new chat and paste the prompt, or use the Open in ChatGPT link above.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.
  • Editor agent

    VS Code

    Microsoft's editor, with whichever agent extension you have installed.

    Setup steps
    1. Sign up for the free trial and request your token from the dashboard; it arrives by email. Set it as MARKETDATA_TOKEN on your machine. The agent cannot create the account or fetch the token for you.
    2. Open the project in VS Code and open the chat panel of your agent extension.
    3. Paste the prompt and send it. The agent reads llms.txt and, if the project does not say, asks whether you want the data in a spreadsheet or in code.
    4. Let it finish with one real call. A stock quote coming back, or a formula filling a cell, means the setup works.

Our llms.txt

What the agent will know

Our llms.txt is written for an agent. These are the rules it reads before it writes a line.

Signing up is your step

Signing up for the trial and requesting your token are your steps. The agent stops, tells you what to do, and continues once MARKETDATA_TOKEN is set.

Two paths, and the SDK comes first

A spreadsheet user gets the Google Sheets add-on and its formulas; a developer starts with the official SDK for Python, JavaScript, Java, C#, Go and PHP. The agent writes no HTTP client of its own while an SDK exists.

The token lives in the environment

MARKETDATA_TOKEN is the variable every Market Data SDK reads. The agent must not ask you to paste the token into chat or write it into source.

The token travels as a Bearer header

In the Authorization header, never as a URL parameter, so it stays out of logs and caches.

A 203 response is a success

It is served from the cache. Code that accepts only a 200 discards valid data.

A 402 and a 403 mean two different things

A 402 means your plan does not cover the request. A 403 means the source IP changed, and the block clears itself after five minutes; a retry loop makes it worse.

Every page has a Markdown twin

Append index.md to the URL and the agent gets the page as text.

Frequently Asked Questions

Yes. Sign up for the free trial; no credit card is needed. Request your token from the dashboard and it arrives by email. Set it in the MARKETDATA_TOKEN environment variable. The agent cannot do any of these steps for you, and the prompt tells it so.

Your agent is ready when you are.

Start the free trial, set MARKETDATA_TOKEN, and paste the prompt.

Start the free trial