News

Starburst opens Context Studio private preview for governed agent context

Starburst’s Context Studio is in private preview, reconciling business definitions and delivering certified context to MCP-compatible AI agents.

D
Nov 18, 2026 · 2 min read

Starburst has opened Context Studio in private preview, giving enterprises a governed store for the business definitions AI agents use to query data. It collects glossary terms, metrics, business rules, synonyms, join keys and data products, surfaces conflicting entries and makes certified context available through the Model Context Protocol (MCP).

Access to the preview runs through a Starburst account team. Galaxy’s documentation says participants need an enrolled Galaxy account and administrator access. Starburst’s Galaxy release notes date private-preview support to Oct. 5. A Starburst product blog describing Context Studio is dated Nov. 18.

The documented Context Studio workflow has three parts: harvest, reconcile and certify. The current Galaxy preview can harvest existing Galaxy data products and query history. A separate Context Studio as Code API lets teams push context maintained in repositories and data pipelines.

Pushed material flows into one context store while retaining the origin of each entity and field. Starburst says the system surfaces disagreements instead of silently choosing one definition, leaving users to resolve open conflicts through the interface or API. Each push represents a connection’s complete current state, so omitting an entity from a later push removes it from that connection’s contribution.

Certification remains more limited in the preview than the broader workflow description suggests. The documentation says all context currently enters the store as certified while Starburst builds out review and certification by named data stewards. Preview users can therefore serve certified context, but they do not yet have the planned steward-approval workflow.

MCP delivery is listed as a current preview function. Teams can point Starburst AIDA, Claude or another MCP-compatible agent at the Context Studio MCP server. Agents can use the stored definitions, metrics, rules and join keys when constructing and running queries. Galaxy applies the individual user’s access policy when each query executes. The release follows other enterprise tools for querying governed data and supplying context to external AI clients.

The roadmap includes out-of-the-box catalog connectors, named-steward workflows, approval history, version history and tracing an answer to the exact context version it used. A Starburst blog post describes steward certification and decision traces more broadly, but the product page and preview documentation do not list those capabilities as available today.

The opened first-party materials also do not specify final pricing, general-availability usage limits, the first catalog connectors or a date for public preview or general availability.

More news