AI in IDE logoAI in IDE

Use Claude in WorkBuddy with a custom model

Connect Tencent's WorkBuddy to AI in IDE as a custom model and use Claude Opus 5.5, Sonnet 5.5, or Haiku 4.5 with one API key.

This page follows WorkBuddy's model configuration guide. It was not run end to end in WorkBuddy; the endpoint below was tested against AI in IDE with CodeBuddy Code, which uses the same OpenAI-format request path.

Setup at a glance

  1. Create the API key. Open API Keys, create a key, and choose the AWSQ or CCMAX channel.
  2. Add a custom model. In WorkBuddy, open Settings → Models, add a model, and choose Custom.
  3. Fill in the fields.
    • URL: https://api.aiinide.com/v1
    • API Key: your dashboard key
    • Model name: claude-sonnet-5-5 (add claude-opus-5-5 and claude-haiku-4-5-20251001 the same way)
  4. Verify. Pick the model in the model selector and send a short prompt. The request appears in Usage.

The URL must end in /v1

WorkBuddy custom models use the OpenAI format and append the standard /chat/completions path. https://api.aiinide.com/v1 becomes /v1/chat/completions, which accepts AWSQ and CCMAX keys. The bare host becomes /chat/completions, the Codex-channel route, and fails with 400 This endpoint requires a CODEX API key.

If you turn on WorkBuddy's custom protocol switch, it skips path completion and calls the URL exactly as written. In that case enter the full https://api.aiinide.com/v1/chat/completions.

A model with a missing URL or API key is greyed out in the selector and reports error 4100.

Models

Model nameChannelUse it for
claude-opus-5-5AWSQ, CCMAXhard refactors, long agent runs
claude-sonnet-5-5AWSQ, CCMAXeveryday coding
claude-haiku-4-5-20251001AWSQ, CCMAXquick edits, cheap checks

WorkBuddy saves custom models, including the API key, only on your machine. Usage is billed to your AI in IDE key, not to WorkBuddy. Rates are on Models and pricing.

FAQ

Which URL does WorkBuddy need for AI in IDE?

https://api.aiinide.com/v1, or https://api.aiinide.com/v1/chat/completions when the custom protocol switch is on.

Can WorkBuddy use the Anthropic endpoint?

WorkBuddy's documentation describes only the OpenAI-format /chat/completions path, so use the OpenAI-compatible endpoint. Claude extensions such as Fast mode and hosted tools are not available through it.

Why can't I select the model in an enterprise workspace?

In WorkBuddy Enterprise, models disabled by an administrator are hidden or greyed out in the selector. Ask the administrator to allow custom models.

Related

Use Claude in WorkBuddy with a custom model | AI in IDE