Configure custom LLM providers in DeepSeek Harness
linziyanleo/dsh-custom-provider · model-adapters
Configure custom LLM providers in DeepSeek Harness.
★ 2GitHub stars
0Forks
2026-09-02Last updated
TypeScriptLanguage
MITLicense
Key features
- A bilingual Web settings page for adding, editing, and removing custom providers.
- When a configured model is used, DSH sends its model request and the resolved API credential to that route's user-configured baseURL using the selected protocol.
- OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages endpoints
- Per-route protocol selection for openai-completions, openai-responses, and anthropic-messages.
- A declarative llm-custom.providers configuration for settings.yaml.
Requirements
- Node.js ^22.19.0 or =24.0.0 (minimum verified version: 22.19.0)
- DeepSeek Harness with the Web profile
Install command
dsh plugin --profile web add github:linziyanleo/dsh-custom-provider