Explosion has not released an MCP server for Prodigy, and we found no community one. The explosion/mcp-server repository this page used to link does not exist. Prodigy's LLM features run through spacy-llm, and its recipes are plain Python, so wrapping a recipe in an MCP tool is possible, but you would be writing that wrapper yourself.
Source and setup verified on 2026-11-03
AI & MLExplosion (official) · ★ 1.4k · updated 2026-10-23
Explosion's library for putting LLM prompts into spaCy pipelines. Prodigy's LLM-assisted annotation recipes are built on it. Not an MCP server.
Best for: LLM pre-annotation inside Prodigy, without MCP.
Explosion (official) · ★ 511 · updated 2024-08-04
Source for recipes such as ner.manual and ner.teach. Each recipe is a Python function, which is the piece an MCP wrapper would call.
Best for: A starting point if you write your own Prodigy MCP wrapper.
Not as of 2026-11-03. Explosion has not published one and we found no community project.
Use Prodigy's built-in LLM recipes, which run on spacy-llm. They pre-annotate examples that you then accept or correct in the Prodigy UI.
Google Gemini server with embedded viewers for generated and edited images, videos, SVG graphics, and landing pages
Official Azure multimodal image-search sample with an embedded viewer for image files returned from Azure AI Search and Blob Storage
AI service marketplace server with an optional embedded run viewer for live job progress and generated image, video, audio, table, and file outputs