AI Agent Hub
Back to plugins
🧩

dsh-oauth-api

admin-security Updated 2026.08.14

Run the following command in DeepSeek Harness:

dsh plugin install hahaha-taotao/dsh-oauth-api

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

Run dsh plugin install hahaha-taotao/dsh-oauth-api in DeepSeek Harness to install this plugin; the source is hosted at https://github.com/hahaha-taotao/dsh-oauth-api .

About this plugin

Juggling API keys for Grok, ChatGPT Codex, Claude Code, and Kimi Code across multiple AI services is tedious and error-prone. dsh-oauth-api, an out-of-tree community plugin for DeepSeek Harness, adds an OAuth login section to Settings so a single browser authorization is all you need—no manual key entry, no juggling of credentials.

The plugin supports four OAuth flows: Device Code for xAI, OpenAI Codex, and Kimi, and a PKCE paste-code#state flow for Claude Code. A successful login never silently changes your default model; you simply select the matching group in the model picker such as xai-oauth, codex-oauth, claude-oauth, or kimi-oauth. All tokens are stored exclusively in local JSON files under $DSH_HOME/oauth/, with nothing sent to any remote store.

This plugin is a good fit if you already run DeepSeek Harness, want OAuth-based access to the four services above, and prefer that credentials stay on your local machine only. Keep in mind it is a community-maintained out-of-tree plugin, not an official DeepSeek product. xAI login requires the api:access scope; if inference complains about a missing scope, simply log in once more.

Use Cases

  • Authorize via browser once to connect Grok, Codex, Claude Code, and Kimi Code without manual API key entry
  • Switch between multiple AI services by simply selecting the matching OAuth group in the model picker
  • Keep all credentials in local JSON files instead of scattering keys across configuration files

Best For

  • Developers already running DeepSeek Harness who want additional model sources
  • Engineers who need unified credential management across multiple AI services to reduce key maintenance overhead
  • Users who prefer fully local credential storage and do not want keys passing through cloud services