Brings the Transkribus REST API into Claude for handwritten text recognition and historical document transcription workflows. You get 301 tools across collections, documents, HTR/OCR recognition, model management, and search, split into eight entry points so you can stay under tool limits. Handles session auth with username/password or direct session ID. Reach for this when you're building workflows around archival research, manuscript digitization, or training custom handwriting models. The split servers let you load only collections, transcription, or admin domains as needed rather than pulling in everything at once.
claude mcp add --transport stdio io.github.lazyants-transkribus uvx transkribus