Can AI agents automate Data API builder?

Microsoft Data API builder (DAB) is automatable today through generated REST and GraphQL APIs. This record is DAB, an open-source MIT engine that exposes configured database entities. It is not the SQL Server engine, Azure SQL as a standalone database identity, or an NL2SQL product. REST lives at /api/{entity} with GET, POST, PUT, PATCH, and DELETE plus OData-style $filter, $select, $orderby, $first, and $after. GraphQL lives at /graphql with matching queries and mutations. Authentication providers documented by Microsoft include Unauthenticated, Microsoft Entra ID, Custom JWT, App Service EasyAuth, Simulator, and On-Behalf-Of. Coverage is the entities in dab-config.json, not the whole database schema. SQL MCP Server ships in DAB 1.7 and later; Microsoft recommends the latest 2.0 release, which is generally available. Tools are DML only: describe_entities, create_record, read_records, update_record, delete_record, execute_entity, and aggregate_records (2.0+). DAB does not support NL2SQL and does not expose DDL. Stored procedures can be registered as named custom MCP tools in 2.0. Transports are streamable HTTP at /mcp and stdio via dab start --mcp-stdio. The official CLI is dab (NuGet Microsoft.DataApiBuilder) for init, add, configure, and start. DAB runs in any cloud or on-premises against SQL Server, Azure SQL, Azure Cosmos DB, PostgreSQL, and MySQL. iPaaS connectors were not opened. Computer-use viability is unassessed. API scores 8 and is the best path. MCP scores 7 because coverage is official and useful but limited to configured entities and DML.

Best path today: api · Overall automatability: 8/10