More~5 min

Connect your AI tools with MCP

Use the General Input MCP to give Claude, ChatGPT, Gemini, or Grok access to every integration, workflow, and skill in your workspace.

The General Input MCP connects your workspace to the AI tool you already use. Once it is paired, your agent can reach every integration you have authorized, run and build workflows and skills, and use the business context behind them, without leaving that chat window.

Why bother

  • Depth. Most native connectors cover the popular actions of an API. General Input documents the whole surface, which works out to roughly three times more tools per integration.
  • One place to manage access. Integrations stay in your General Input vault. Connect a service once and every AI tool you pair can use it.
  • Credentials stay yours. Secrets are injected at runtime inside the sandbox and stripped before anything returns to the model.

Get your connector details

  1. Open Settings and find the MCP section.
  2. Copy the Server URL and the Connector name. You will paste both into whichever AI tool you are connecting.

Connect Claude

  1. In settings, click Manage connectors. That opens your Claude settings page.

  2. Go to Connectors and click Add, then Add custom connector.

  3. Paste the connector name and the server URL, then click Continue, scroll down, and click Add.

  4. Authenticate: click Connect, then Allow. This pairs that workspace to your Claude account.

Connect other tools

The same server URL works with anything that speaks MCP. The MCP section in settings has copy-paste steps for:

  • Claude on the web and in the desktop app
  • ChatGPT desktop app
  • Claude Code as a single command
  • Codex CLI

Gemini, Grok, and other MCP clients use the same server URL and connector name.

Try it

Ask for something that needs more than one system. A good first prompt:

Using the General Input MCP, pull everything you can find on our largest customer and summarize where the account stands.

The agent searches your connected integrations, calls the operations it needs, and answers from live data rather than guesswork.

Save what works

When the agent produces something you will want again, have it write the process back into your workspace:

Now that you understand how to build this, create a skill in General Input with the template and process.

The skill lives in your workspace from then on, so the next run is consistent instead of improvised.

Related