DataHub

DataHub is a metadata platform. It catalogs the data assets across your stack — along with their lineage, ownership, and glossary terms — and stores the context pages that document them. Honeydew’s AI can read those context pages as context when answering questions.

​

Two Context Layers

Honeydew and DataHub cover different levels of the stack: The two are complementary. DataHub describes the assets and documents them in context pages; Honeydew defines the logic that turns them into governed metrics, and generates the SQL. Reading DataHub context pages into Honeydew avoids restating business knowledge that is already written down in the catalog.

​

DataHub as a Context Source

DataHub is supported as an External Knowledge source, so the AI can fetch context pages from DataHub over MCP. Context pages are the documents DataHub stores alongside the catalog — knowledge articles, runbooks, decisions, and FAQs. They are the DataHub equivalent of a Notion or Confluence page: prose written for readers, rather than catalog metadata about an asset. Use this to reuse business knowledge that DataHub already governs, instead of restating it as Honeydew context. Definitions stay owned in one place, and Honeydew reads the current version at query time.

​

Setup

1. Go to Settings → MCP connections and connect DataHub. Honeydew connects to DataHub Cloud’s hosted MCP server at https://mcp.datahub.com/mcp over OAuth. Each user authenticates individually, so content is fetched with their own permissions.
2. Create an External Knowledge context item whose external_source.tool is mcp__datahub__search_documents and whose resource_id is the URN of the context page to fetch.

DataHub also exposes mcp__datahub__grep_documents, which searches within context page content using a regular expression. See External Knowledge for the full field reference.