Argorant Help
EN|DEargorant.comOpen Argorant
Help/AI agents & API

Use Argorant in ChatGPT without a connector

Published by Argorant

When you cannot add a connector: a custom GPT action or a terminal that uses your private key, set up safely.

The easiest way to use Argorant in ChatGPT is the connector, described in Set up Argorant in ChatGPT with a connector. This page is for the cases where you cannot add one, or where you want something the connector does not give you, such as a shared custom GPT with your ICP rules built in. Those setups use a private API key.

Use the OAuth MCP connection when your client supports custom remote MCP servers. Otherwise use the CLI from a Node/npm-enabled terminal, or REST from an authorized HTTP environment. Pasting a personal key into a chat does not create a connection or bypass the client's restrictions.

When a key is the right choice

SituationUse
You can add a custom MCP server in ChatGPTThe connector. No key needed.
You build a custom GPT with actions for your teamA key stored in the action's authentication
Your AI has a terminal (Codex, Claude Code, a server)The command line or REST with a key in the environment
A plain chat with no toolsNeither works. A key pasted into the chat connects nothing.

Who can do it

Anyone with an Argorant account and a confirmed email address can create a key. Building a custom GPT with actions depends on your ChatGPT plan and workspace settings. A key acts with your account's access and spends from your workspace's balance, so whoever can use the key can spend your credits.

Create a key

  1. Open Settings, then API keys.
  2. Under New key, type a name that says where it is used, such as chatgpt-gpt-action.
  3. Click Create key. The key starts with ag_live_ and is shown only once. Click Copy and put it straight into the place it will live.
Settings, API keys, with the Key name field and the Create key button under Your keys
Settings, API keys. Name the key after where it is used, then click Create key.

You can also click Create private key on the CLI or REST API tab of AI connections. It does the same and fills the key into the commands on that tab.

A custom GPT with an action

  1. In the GPT editor, add an action and import the schema from https://docs.argorant.com/openapi.json. Keep only the operations the GPT needs, for example the account check, the people count and the people preview.
  2. Set the authentication to an API key sent as a Bearer token, and paste your key there. It is stored with the action, not in the conversation.
  3. Write your rules into the GPT's instructions: who you sell to, which filters to use, and that it must ask before any reveal or export.
  4. Test it with an account check. A real answer shows your account and credits. An answer without a call to Argorant is not a test.

Every person who uses that GPT spends from the account that owns the key. For a shared GPT, keep it to free operations (account, count, preview) unless you want the team to spend credits through it.

The two facts any HTTP tool needs

  • The base URL is https://app.argorant.com. The API reference is at docs.argorant.com.
  • Send the key as Authorization: Bearer ag_live_.... Never put it in a URL.
curl --fail-with-body https://app.argorant.com/api/mcp/account \
  -H "Authorization: Bearer $ARGORANT_API_KEY"

A terminal instead

If your AI runs commands, the command line is simpler than raw HTTP. Set the key in the environment and run:

npx argorant login "$ARGORANT_API_KEY"
npx argorant whoami

Then ask it for counts and previews. Note that reveals and exports from the command line do not stop to ask when they run inside an agent. They spend at once.

Prompts to try in your GPT

Check my Argorant account.

Calls the account endpoint. Shows the account, credits left and today's usage. Free.

How many marketing directors at software companies in France?

Calls the people count. Free. During the trial and without a plan, counts through a key are limited to 10 a minute and 500 a day per account.

Show five masked examples for that search.

Calls the preview. Names are masked, no emails. Free.

Reveal 5 of them, but tell me the cost first.

Calls the reveal if your action includes it. 1 credit per person whose email is new to your account. The GPT has to ask you first because you told it to in its instructions. The API does not ask on its own.

Keep the key safe

  • One key per use. If a GPT or a script leaks, you revoke only that key.
  • Never paste a key into a chat message, a shared document or a URL.
  • Revoke a key under Settings, then API keys, with Revoke. It stops working at once. The next request fails with 401.

Common mistakes

  • The key in the chat. ChatGPT cannot use it from there, and the key is now in your chat history. Revoke it and create a new one.
  • The key in the URL. Requests fail and the key can end up in logs. Use the Authorization header.
  • 401 Unauthorized. The key is wrong, revoked, or the header is missing "Bearer ".
  • 402 Payment Required. The action needs credits or a plan. The response says what is missing.
  • 429 Too Many Requests. You hit the speed limit. Wait and try again.

What it costs

Search, counts and masked previews do not deduct credits. Account permissions and rate limits still apply. New accounts get a 7-day trial after email confirmation: 100 credits and 300 campaign emails, no automatic subscription; when it ends, campaigns pause and unused trial credits expire until a plan is added. The app, REST API, CLI and MCP share the same account and balance; connecting another client or creating an API key does not add a second grant. Existing balances and subscriptions are unchanged.

What happens next

Everything the key does shows in your account, under Lists, Exports and your credit balance. When the work is done, revoke the key. For more on keys see key rotation for agents and 401 and 403 errors.

Related
Set up Argorant in ChatGPT with a connectorKey rotation for agents401 and 403: authentication and scope errorsAPI keys and the REST API

Still stuck? support@argorant.com

Help homeargorant.comPricingRoadmap
support@argorant.com