Summary
One prompt out, one answer back, with nothing per-call to decide.
The single parameter is originPrompt, so there is no model selection, no temperature and no message history to assemble — the prompt you already have goes across as it stands. Everything that would otherwise be a per-call option lives in the environment instead, which makes this a small and easily replaced piece rather than a configurable one.
What it is
A single-tool bridge. get-deepseek-thinker takes an originPrompt and returns DeepSeek's response; BASE_URL decides which endpoint answers and API_KEY authenticates against it.
What you get
- DeepSeek's response to a prompt handed over as originPrompt, unmodified
- An endpoint you choose through BASE_URL rather than one fixed in the code
Requirements
An endpoint to call and a credential for it: BASE_URL and API_KEY. Node on your PATH.
Setup effort
One command plus a key — npx -y deepseek-thinker-mcp, then supply credentials