Schema explorer

Introspect runs the standard introspection query and loads the schema (Schema shows or hides the explorer):

• Autocomplete for fields, arguments, types, directives and variables (Ctrl+Space).
• Validation as you type, with errors underlined in the editor.
• Hover documentation for types.
• Explorer listing root types, all types (objects, interfaces, unions, enums, inputs, scalars) and directives. Click a type to drill down and a field to insert it. The SDL toggle shows the printed schema.

If introspection is disabled on the server, the error explains why and suggests alternatives.

Build an operation

On the root types (Query, Mutation, Subscription), each field has a Build button. It writes a complete operation for that field into the editor:

• a variable for each argument, declared with its type, and placeholder values in Variables ("" for strings and IDs, 0 for numbers, false, the first enum value, and input objects with their required fields);
• a selection of the field's scalar fields, and of nested objects two levels deep. Fields that need arguments, deprecated fields and types already on the path (cycles) are left out; unions get an inline fragment per type.

If the editor already has a query, you're asked before it is replaced. Fill in the variables and Run.

The same builder is in the terminal and for AI agents:

testpion graphql-op Query.patient --endpoint http://127.0.0.1:4011/graphql
testpion graphql-op addPatient --schema schema.graphql --depth 1 --json

AI agents use the graphql_operation MCP tool, which introspects the endpoint and returns { operation, operationName, query, variables }, so they don't have to guess field names.