Connecting to LLMs: ChatGPT, Gemini, and Proprietary models

Overview: the Danomics MCP server

Any AI assistant that supports remote MCP can work with your Danomics data by connecting to one URL: https://app.danomics.com/mcp. The three articles below cover ChatGPT, Google Gemini (Gemini CLI and Antigravity), and any other MCP-compatible client.

What you can do once connected. Ask the assistant to find wells, read LAS logs and formation tops, check petrophysical interpretation (CPI) results, run flows and decline-curve jobs, and manage files in your Danomics workspace. The assistant acts with your Danomics permissions only. It sees what you can see and nothing more.

Before you start

• An active Danomics account. Sign-in happens in your browser the first time you connect.
• An AI client that supports remote MCP servers over HTTPS (Streamable HTTP transport) with OAuth sign-in.
• For company-managed AI accounts (ChatGPT Business/Enterprise, Google Workspace), your IT admin may need to allow custom MCP connections first.

Article 1: Connect ChatGPT to Danomics

You add Danomics to ChatGPT as a custom app (ChatGPT's name for a connector) using developer mode, on ChatGPT web. It takes about five minutes.

Plan requirements (OpenAI help)

Custom apps work on ChatGPT web only, not the mobile apps. Agent mode does not use custom apps; deep research can use them for read actions only.

Step 1: Turn on developer mode

1. Business / Enterprise admins: go to Workspace settings → Permissions & Roles → Connected Data and enable developer mode (Enterprise/Edu can grant it to specific groups).
2. Each user with access: open Settings → Apps → Advanced settings and toggle Developer mode on.

Step 2: Create the Danomics app

1. Go to Settings → Apps → Create (admins can also use Workspace settings → Apps → Create).
2. Enter:
   • Name: Danomics
   • Description: Danomics well, log, petrophysics and workflow data
   • MCP server URL: https://app.danomics.com/mcp
   • Authentication: OAuth
3. Click Scan Tools. A Danomics sign-in window opens. Sign in and approve access.
4. When the tool list appears, click Create. The app shows under Settings → Apps with a Dev label.

Step 3: Use Danomics in a chat

1. Start a new chat, open the + / tools menu in the composer, and select Danomics (or @mention it).
2. Try: “List the well databases in my Danomics workspace.”
3. App selection applies to one message. @mention Danomics again when a follow-up needs fresh data.
4. ChatGPT asks you to confirm before write actions such as saving tops edits or starting a job. Review each one.

Step 4 (Business / Enterprise): Publish to your team

An admin opens Workspace settings → Apps → Drafts, selects Danomics and clicks Publish. Enterprise admins can use Configure Actions to allow only read tools and Configure Access to limit it to groups. Teammates then find Danomics in their Apps list, labelled custom, and each signs in with their own Danomics account.

ChatGPT freezes the tool list at publish time. When Danomics adds or changes tools, Enterprise admins click Refresh under the app's Action control; Business admins must recreate and republish the app.

ChatGPT troubleshooting

• No Create button or Apps option: developer mode is off, or your plan or admin doesn't allow it.
• Only search/fetch tools appear: developer mode is not actually enabled for your account.
• Danomics stops responding after a while: reconnect from Settings → Apps → Danomics.
• Tool calls fail after a Danomics update: ask your admin to refresh or republish the app.

Article 2: Connect Gemini CLI and Google Antigravity to Danomics

Google's developer tools, Gemini CLI and the Antigravity IDE, both connect to Danomics through a small JSON config entry and a one-time browser sign-in. They use different config files and different URL keys, so copy the block for your tool exactly.

The consumer Gemini app (gemini.google.com) does not currently accept custom MCP servers. Use Gemini CLI or Antigravity.

serverUrl

Gemini CLI

1. Add the server with one command:

gemini mcp add --transport http --scope user danomics https://app.danomics.com/mcp

1. Or add it to ~/.gemini/settings.json by hand:

{
"mcpServers": {
"danomics": {
"httpUrl": "https://app.danomics.com/mcp",
"oauth": { "enabled": true },
"timeout": 60000
}
}
}

1. Start gemini, then run /mcp auth danomics. A browser window opens; sign in to Danomics and approve access. Gemini CLI stores and refreshes the token for you.
2. Run /mcp to confirm danomics shows as connected and lists its tools.
3. Try: “Using Danomics, list my well databases.” Gemini asks before each tool call; approve it.

Use httpUrl, not url. In Gemini CLI, url means the older SSE transport and the connection will fail with what looks like a network error (source). OAuth sign-in needs a local browser, so it will not complete over plain SSH or in a headless container.

Google Antigravity

1. In the Agent panel, open the … menu → MCP Servers → Manage MCP Servers → View raw config. This opens mcp_config.json.
2. Add Danomics:

{
"mcpServers": {
"danomics": {
"serverUrl": "https://app.danomics.com/mcp"
}
}
}

1. Save, then refresh the MCP server list (or restart Antigravity). Complete the Danomics sign-in when prompted.
2. Confirm Danomics appears with its tools listed, then ask the agent a Danomics question.

Use serverUrl. Antigravity rejects url and httpUrl, so a block copied from Gemini CLI or Cursor will not connect until the key is renamed (source). To pause Danomics without deleting it, add "disabled": true; to hide specific tools, list them in "disabledTools".

Article 3: Connect any MCP-compatible client to Danomics

If your AI tool supports remote MCP servers, it can connect to Danomics with the URL https://app.danomics.com/mcp and OAuth sign-in. The only thing that changes between tools is where the config lives and what the URL key is called.

The general recipe

1. Find your tool's MCP settings (a settings screen, or a JSON/TOML config file).
2. Add a remote server named danomics, transport HTTP / Streamable HTTP, URL https://app.danomics.com/mcp.
3. Leave API keys and custom headers blank. Choose OAuth if asked.
4. Restart the tool or start a new session. Most clients read their server list only at startup.
5. Complete the Danomics sign-in in your browser.
6. Confirm the Danomics tools appear, then ask a test question such as “List my Danomics well databases.”

Config for common clients

Troubleshooting and security (all clients)

Security notes

• The assistant uses your Danomics permissions. It cannot see data you can't see in the Danomics app.
• Each person signs in with their own Danomics account. Never share tokens or put credentials in shared config files.
• Review write actions (saving tops edits, changing configs, deleting files, starting jobs) before approving them. Avoid “always allow” or --trust settings for write tools.
• Data the assistant retrieves is sent to that AI provider under your organisation's agreement with them. Check your company policy before connecting.
• To disconnect, remove Danomics from the client's apps/connector list or config file.

Sources

• OpenAI Help: Developer mode and MCP apps in ChatGPT
• Copyleaks MCP client setup (Gemini CLI, Antigravity, Codex, Cursor config conventions)