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

# VS Code Chat

> Connect VS Code built-in Chat to a qualified Planck model.

In Planck, open **Connect your coding tools**, select **VS Code Chat**, and choose
an available qualified model. VS Code is the editor; Codex and Claude extensions
have separate setup instructions.

1. Open **Manage Language Models → Add Models → Custom Endpoint**.
2. Enter the Planck credential in the API-key prompt, using VS Code secret storage.
3. Merge Planck's generated model configuration into `chatLanguageModels.json`.
   Keep the `${input:planckApiKey}` reference rather than putting a raw key in JSON.
4. Use the generated full endpoint URL and API type. The configuration includes
   model limits, tool support, bearer authentication, and a temporary setup header.
5. Select the model in Chat, start a new conversation, and send the displayed test prompt.
6. Wait for Planck to show the matched completed request. Remove the setup header
   after verification, preserving authentication and model settings.

For remote development, configure the VS Code host that runs the model provider.
A successful browser gateway test does not verify this editor connection.

See [VS Code's Custom Endpoint reference](https://code.visualstudio.com/docs/agent-customization/language-models).


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