SYSTEM Cited by 2 sources

DataHub

DataHub (datahubproject.io) is an open-source metadata platform — a data catalog that tracks tables, columns, ownership, lineage, glossary terms, and quality signals across a company's data assets. Originally developed at LinkedIn and now a community-maintained project.

Role in this wiki

DataHub appears in the corpus primarily as the substrate of PinCat — Pinterest's internal data catalog. Pinterest layers tier tags, owners, retention policies, and column-level glossary terms on top of DataHub, and PinCat becomes the system of record that drives the Analytics Agent's schema grounding + governance-aware ranking.

(Source: sources/2026-03-06-pinterest-unified-context-intent-embeddings-for-scalable-text-to-sql.)

Core concepts (as used by Pinterest)

• Table tier tags — Tier 1 / 2 / 3 classification of production quality (data-governance-tiering).
• Glossary terms — reusable business concepts (e.g. user_id, pin_id) that unify different column names across tables. DataHub docs: business-glossary.
• Ownership + retention policies — metadata that gates deprecation decisions and drives AI ranking trust signals.

Seen in

• sources/2026-03-06-pinterest-unified-context-intent-embeddings-for-scalable-text-to-sql — Pinterest's PinCat is built on DataHub.
• sources/2026-05-28-cloudflare-how-we-built-cloudflares-data-platform-and-an-ai-agent-on-top-of-it — DataHub is the metadata catalog of Cloudflare Town Lake. "Every table, column, owner, lineage edge, and glossary term lives there." Skipper queries DataHub for Layer 1 (schema + usage metadata, including historical join-pattern data) and Layer 2 (human-written table descriptions
• glossary + curated tags) of its five-layer grounded context. Distinctively, DataHub is also the destination for .meta.json documentation emitted by the Transformer ELT engine on every successful run — making DataHub the substrate of code as context for data agents (Skipper's Layer 3).

Related

• systems/pinterest-pincat
• systems/cloudflare-town-lake — second canonical wiki instance: DataHub as the metadata catalog of Town Lake.
• systems/cloudflare-skipper — queries DataHub for Layers 1 + 2 + 3 of its grounded context.
• systems/cloudflare-transformer-elt — emits per-node .meta.json to DataHub on every successful run.
• data-governance-tiering
• glossary-term-propagation
• code-as-context-for-data-agents — DataHub as the storage substrate for code-derived semantic context.
• layered-grounded-context-for-data-agent — Skipper's five-layer model where DataHub powers three of the five.