Summary
A paper in a language you do not read stops being a dead end.
Search and translation sit behind one key, so identifying a relevant paper and queueing its translation happen without leaving the session. The translation path is deliberately asynchronous — create_translation hands back a task, get_translation polls it for status and result URLs — which means the client has to come back rather than block. optimize_math_formula is the option worth setting: it exists for documents where the notation is the content.
What it is
A client for Suppr's literature service: semantic search over academic papers and an asynchronous document translation queue.
What you get
- Semantic search over academic literature from a natural-language query, with a topk cap and an auto_select option
- Translation jobs created from a local file path or a URL, between a source and target language, with optimize_math_formula for notation-heavy documents
- Job status and result URLs retrieved by task_id, since translation runs asynchronously rather than returning in the call
- A paginated list of every translation job you have queued before
Requirements
A Suppr API key in SUPPR_API_KEY, and a local file path or a reachable URL for anything you want translated.
Setup effort
One command plus a key — npx suppr-mcp, then supply credentials