AI Agent Hub
Back to plugins
🖥️

dsh-trae-api

Client Updated 2026.08.26

Run the following command in DeepSeek Harness:

dsh plugin install Casually/dsh-trae-api

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Casually/dsh-trae-api in a DeepSeek Harness environment to install the plugin; the source repository is https://github.com/Casually/dsh-trae-api .

About this plugin

You have Trae Work CN credits sitting idle, while Claude Code, Cursor, Cline, and Windsurf all demand an OpenAI or Anthropic endpoint. dsh-trae-api fills that gap. Running as a DeepSeek Harness plugin, it automatically decrypts authentication tokens from the local Trae IDE storage, spins up a local proxy that speaks both the OpenAI and Anthropic protocols, and lets your existing credits power any third-party AI coding tool without a single additional API invoice.

The proxy covers the full programming loop: three endpoint families (/v1/chat/completions, /v1/messages, /v1/responses), streaming SSE output, Claude Code tool-calling support, and automatic model-name mapping (send claude-sonnet-4-6 and GLM-5.2 is dispatched; send gpt-4o and DeepSeek-V4-Pro takes the call). Tokens refresh and persist on their own, and the context window is truncated to stay within limits. Since v1.1.0 a DSH Web settings page offers live status, per-endpoint request statistics, one-click token refresh, upstream connectivity probing, and hot-reload of port, API key, and context limits the moment you save.

Who is this for? Developers who already hold credits in Trae CN, TRAE SOLO, or the international editions and use at least one of Claude Code, Cursor, Cline, or Windsurf, and would rather put those credits to work than open another API subscription. The proxy binds to 127.0.0.1 by default, enforces API-key authentication out of the box, and keeps the security perimeter tight.

Use Cases

  • Call GLM, DeepSeek, Qwen, and other models behind Trae credits from Claude Code or Cursor
  • Code in Cline or Windsurf through a local OpenAI-compatible endpoint with zero extra API spending
  • Route one set of Trae credits into multiple AI coding tools through a single localhost proxy

Best For

  • Developers holding Trae Work CN credits who rely on AI coding tools daily
  • Full-stack engineers seeking unified multi-model access with lower API costs
  • Individual developers who prefer local deployment over per-tool subscriptions