MPS Language Aspects — Overview
SkillDev toolsEntry point for authoring or modifying an MPS language — index of all aspects (structure, editor, constraints, behavior, typesystem, intentions, actions, generator, textgen, accessories, generation plans), the recommended authoring order, cross-aspect call patterns, dependency/rebuild rules, and validation flow. Start here when defining a new language or when you do not know which aspect-specific skill applies.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the MPS Language Aspects — Overview skill
What this skill tells your AI
The instructions your AI receives, as published by jetbrains/mps in .agents/skills/mps-language-aspects-overview/SKILL.md and read by ahel’s review.
An MPS language is a collection of aspects, each a separate model inside the language module. Together they define syntax, semantics, UI, and output. Edit each aspect through its own language (its DSL). This skill is the router; each aspect links to a dedicated skill.
Critical Directives
- Always prefer MPS MCP tools over hand-editing
.mpsXML. Hand edits can silently corrupt model files. - Keep aspect rules thin and declarative — push logic into behavior methods, then call them from any aspect.
- When something in a consumer is wrong, re-check the root cause on the language side (structure / generator / typesystem) before patching generated Java.
- Choose a modularity strategy before authoring when this language must relate to others — reference, extend, embed, or adapt them. The kind you pick (referencing / extension / reuse / embedding) changes how you use structure, constraints, typesystem, and generator. See
mps-language-modularity.
Aspects and Where to Find Detailed Guidance
| Aspect | Aspect ID / Stereotype | Model file | Language DSL | Detailed skill |
|---|---|---|---|---|
| Structure | structure | models/structure.mps | jetbrains.mps.lang.structure | mps-aspect-structure-concepts |
| Editor | editor | models/editor.mps | jetbrains.mps.lang.editor | mps-aspect-editor, mps-aspect-editor-menus-and-keymaps |
| Constraints | constraints | models/constraints.mps | jetbrains.mps.lang.constraints | mps-aspect-constraints |
| Behavior | behavior | models/behavior.mps | jetbrains.mps.lang.behavior | mps-aspect-behavior |
| Typesystem & Checking | typesystem | models/typesystem.mps | jetbrains.mps.lang.typesystem | mps-aspect-typesystem |
| Intentions | intentions | models/intentions.mps | jetbrains.mps.lang.intentions | mps-aspect-intentions |
| Actions | actions | models/actions.mps | jetbrains.mps.lang.actions | mps-aspect-actions |
| Generator | @generator (stereotype) | generator/template/main@generator.mps | jetbrains.mps.lang.generator | mps-aspect-generator |
| TextGen | textGen | models/textGen.mps | jetbrains.mps.lang.textGen | mps-aspect-textgen |
| Dataflow | dataFlow (camelCase) | models/dataFlow.mps | jetbrains.mps.lang.dataFlow | mps-aspect-dataflow |
| Migrations | migration | models/migration.mps | jetbrains.mps.lang.migration | mps-aspect-migrations |
| Enhancement scripts | scripts | models/scripts.mps | jetbrains.mps.lang.script | mps-aspect-migrations |
| Find Usages | findUsages (camelCase) | models/findUsages.mps | jetbrains.mps.lang.findUsages | — |
| Refactorings | refactorings | models/refactorings.mps | jetbrains.mps.lang.refactoring | — |
| IDE plugin | plugin | models/plugin.mps | jetbrains.mps.lang.plugin | mps-ide-plugin |
| Feedback | feedback | models/feedback.mps | jetbrains.mps.lang.feedback | — |
| Accessories / Dependencies | — (no model created) | .mpl and per-model | — | mps-aspect-accessories |
| Code inside bodies | — | any aspect that holds BL | baseLanguage, smodel, collections, closures | mps-model-manipulation |
| Inline node literals | — | any aspect using quotations | jetbrains.mps.lang.quotation | mps-quotations |
Aspect IDs are case-sensitive and create models named <langModule>.<aspectId> with no @ suffix. Stereotypes (@generator, @genplan, @tests, @descriptor) are a separate mechanism for non-aspect models. Authoritative table with usage notes and pitfalls: aspect-model-stereotypes.md.
Additional aspects exist (Find Usages, Refactorings, Scopes Provider, Plugin). They follow the same "per-concept rule root + body" pattern — consult companion skills in .agents/skills/ or the MPS user guide when needed.
Generation Plans (cross-cutting, not an aspect of a single language) are covered by mps-aspect-generation-plan. Use when you need to control generator ordering, enable cross-model reference resolution with checkpoints, fork generation into parallel branches, or conditionally contribute plans via PlanContribution.
Typical Authoring Order
- Structure — concepts, properties, children, references, inheritance.
- Editor — notation, cell layouts, projections; iterate alongside sample models.
- Constraints — property validators, reference scopes, placement rules.
- Behavior — concept methods/constructors; callable from any aspect.
- Typesystem & checking — types, subtyping, semantic errors with quick fixes.
- Intentions — alt-enter refactorings and shortcuts.
- Generator — transform to a target language (often BaseLanguage).
- TextGen — only if the target is plain text the language itself emits.
- Accessories / runtime / dependencies — wire into the
.mpl, ensure consumers can find everything.
Most iteration is circular: structure ↔ editor ↔ constraints ↔ typesystem.
Cross-Aspect Calls
- Any aspect body can call behavior methods on nodes (
node.m(...)). - Typesystem rules call quick-fix intentions via
error ... fix { ... }or by referencingQuickFixroots. - Generator templates call behavior methods for nontrivial logic and use mapping labels for cross-template references.
- Constraints may call behavior methods for reusable checks.
- Editors may call behavior methods in query/if/cell-model blocks.
Keep logic in behavior methods; keep aspect rules thin and declarative.
MCP Tools You'll Use Often
- Exploration:
mps_mcp_get_project_structure,mps_mcp_search_concepts,mps_mcp_get_concept_details,mps_mcp_print_node. - Authoring:
mps_mcp_create_root_node,mps_mcp_insert_root_node_from_json,mps_mcp_update_node,mps_mcp_parse_java_and_insert. - Structure-specific:
mps_mcp_alter_structure(CREATE_CONCEPTS, CREATE_ENUM, UPDATE_CONCEPT_*),mps_mcp_query_structure(inheritance/aspect queries); example nodes of a concept viamps_mcp_query_nodes(FIND_INSTANCES). - Modules/models:
mps_mcp_create_module,mps_mcp_create_model,mps_mcp_model_dependency,mps_mcp_model_used_language,mps_mcp_module_dependency. - Validation:
mps_mcp_check_root_node_problems, language/module build via IDE MCP.
Dependencies and Rebuilds
After adding/changing:
- Concepts (structure) — downstream aspects' references by concept may break; validate them. Rebuild language.
- Behavior method signatures — all callers (generator, typesystem, constraints, intentions, editor queries) must be re-validated. Rebuild language.
- Generator templates — rebuild generator module (it's separate from the language module).
- Runtime solutions — rebuild the runtime solution; consumers pick up changes on next regeneration or reload.
- After large changes:
mps_mcp_reload_allto refresh MPS's runtime concept registry.
Validation Flow
mps_mcp_check_root_node_problemson edited roots / aspect models — catches concept-level problems.- Build the language module (IDE MCP) — catches compile errors in generated descriptors.
- Regenerate a sample model — catches generator/textgen issues.
- Inspect sample in MPS UI — catches editor / usability issues.
When something in a consumer is wrong, always re-check the root cause on the language side (structure / generator / typesystem) before patching generated Java.
Signals
- GitHub stars
- 2k
- Forks
- 309
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
mps-language-aspects-overview- Source
- github.com/jetbrains/mps