Introduction to the DeepSeek Harness Plugin dsh-llm-ai
`dsh-llm-ai` is a multi-provider LLM adapter plugin for the DeepSeek Harness (DSH) platform, replacing the older `dsh-llm-pi-ai`. It operates based on the `openai-completions` protocol, supports automatic retrieval of model and provider information from the models.dev registry, and provides features such as offline startup, streaming, stable error codes, dynamic configuration, and credential management. The plugin is installed via the CLI, automatically disables the older plugin to avoid conflicts, and supports hot configuration updates. It is suitable for DSH users who need to connect to multiple OpenAI-compatible models.
Read Moredsh-plugin-devin-bridge: Connect Devin hosted models to the DSH model provider system
dsh-plugin-devin-bridge is an MIT-licensed plugin maintained by Arborsm, designed to adapt Devin's Connect RPC API to the dsh LLM interface. By registering a `devin` provider, it resolves format discrepancies between Devin models (e.g., glm-5.2, swe-1.7) and dsh's local protocols (e.g., GenerateOptions, StreamChunk), implementing streaming response decoding, tool call mapping, and image input support. The plugin features a dynamic model directory, supports 5-minute TTL caching, and gracefully falls back to static configuration on network failures. Installation requires Node 22.19+ and dsh 0.1.0-rc.6+, and authentication is required.
Read More