Wraps 65 spatial transcriptomics methods into 20 schema-validated MCP tools so LLMs can orchestrate reproducible workflows instead of generating arbitrary analysis scripts. Covers the standard pipeline: load 10x Visium or Xenium data, identify spatial domains with SpaGCN or STAGATE, deconvolve with Cell2location or RCTD, run cell-cell communication via LIANA, annotate with Tangram or scANVI, and compute spatial statistics like Moran's I or Ripley's K. Tool parameters constrain method selection, turning natural language prompts into deterministic calls. Particularly dense for multi-sample integration (Harmony, scVI), trajectory inference (CellRank, Palantir), and spatially variable gene detection (SpatialDE, SPARK-X). Designed for researchers who want conversational access to ScanPy-adjacent tooling without writing Python notebooks.
claude mcp add --transport stdio cafferychen777-chatspatial uvx chatspatial