Reference

Runtime API reference: the Graphitron.buildSchema(…​) facade (single-arg and federation two-arg overloads), the Graphitron.newExecutionInput(…​) schema-driven factory (DSLContext first, then typed contextArgument slots reflected from your @service / @condition directives), the sealed GraphitronContext per-request interface (getDslContext, getContextArgument, getValidator), and the NodeIdEncoder utility (generic encode/peekTypeId/decodeValues plus per-Node encode<TypeName> and decode<TypeName> helpers).

Maven plugin configuration: every parameter on every goal of graphitron-maven-plugin. Covers the three goals (generate, validate, dev), the shared <configuration> parameters (outputDirectory, outputPackage, jooqPackage, schemaInputs, namedReferences), the dev-only watch parameters (port, debounceMs, skipInitial), and the two complex POM bindings (<schemaInput>, <namedReference>).

LSP requirements: the LSP’s native runtime. Both the GraphQL grammar and the libtree-sitter runtime ship bundled, so there is no system dependency to install. Covers the supported host architectures and the startup diagnostics.

LSP inlay hints and classification hover: the three editor-side config keys (graphitron.inlayHints.inferredDirectives, graphitron.inlayHints.classification, graphitron.hover.classification) that surface Graphitron’s inference and classification layers inside the editor. Default off; opt in per key.

Diagnostics glossary: every closed-set diagnostic code the validator emits, organised by category. Covers the three top-level rejection kinds (author-error, invalid-schema, deferred), the nine name-lookup attempt kinds (column, table, foreign-key, service-method, lifter-method, enum-constant, type-name, nodeid-key-column, dml-kind), and the four deferred emit-block reasons.

Deprecations: aggregator for every deprecated directive argument, input field, and whole directive in the rewrite, with migration guidance pointing back at the canonical directive page.