What it may solve
DeepSeek Harness (dsh) plugin to compare LLM API cost across models and providers: prices the current session on the model that ran it, then reprices the same token counts against 15 routes on file. DeepSeek peak/off-peak resolved per UTC hour, cache read and write priced separately, every rate with its source and check date.
Imported third-party catalog description; not a Registry verification conclusion.