dsh-sensenova-provider

DeepSeek Harness (DSH) centers its provider ecosystem on official channels, so plugging in an OpenAI-compatible service like SenseNova usually means hand-rolling request scripts or retyping environment variables after every restart. dsh-sensenova-provider collapses that into a single standard sensenova provider route you select directly from the Models page, eliminating the repetitive integration work.

The headline feature is multi-account API-key rotation over one shared base URL. Configure several keys and the plugin will automatically cool down a key on 429 (honoring Retry-After with a 60-second fallback), disable a key on 401, and surface the earliest reset time when all keys are exhausted. It also fetches a live model catalog, streams responses via SSE, and ships a dedicated Web settings section. All credentials are stored through the DSH credentials service and are never logged or forwarded to the model directly.

If you are running DSH 0.1.2-alpha or later and want SenseNova folded into your unified LLM routing, need load-balancing across multiple keys, or simply want to stop maintaining environment variables by hand, this lightweight plugin is built for exactly that.