Create or refresh project-local DSL skills under `.agents/skills/<dsl-name>-dsl/` for one or more MPS languages after discovering concepts, sandbox examples, JSON blueprints, references, or gotchas.
84%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
—
The risk profile of this skill
Companion names in this skill are lazy dependencies: load only those relevant to the current task. If this skill came from an MCP server, use the host's skill loader to resolve the companion's unique discovered entry URI on the same host-assigned originating server. If the host has no server-backed skill loader, stop and report that limitation; do not silently fall back to a filesystem copy. If this skill came from a filesystem catalog, load the named sibling from that same catalog at <skills-root>/<skill-name>/SKILL.md, even if remote skill loaders are also available. Do not invent a tool name or server endpoint.
.agents/skills/<dsl-name>-dsl/, or the existing one is outdated..mps XML.mps_mcp_get_project_structure with includeModels: false. Identify all editable DSL language modules and derive each concept-tools language ref as l:<uuid>:<languageName>.mps_mcp_get_project_structure with the relevant sandbox/example solution as startingPoint and includeRootNodes: true. Record editable model refs and representative root refs.mps_mcp_get_concept_details with the relevant l:<uuid>:<languageName> refs. Capture rootable concepts, concrete children, properties, child roles, references, and useful shortDescription text.mps_mcp_print_node only on representative roots or subtrees needed for reference targets, required roles, or reusable blueprints, and derive blueprints from that output rather than writing them by hand. Avoid dumping every root..agents/skills/<dsl-name>-dsl/ for each selected scope. Preserve user-added notes unless they are stale or wrong. In every generated file, write the temp directory as $TMPDIR (%TEMP% on Windows), never as the resolved directory (/private/var/folders/…/T, C:\Users\…\Temp), which differs per machine..DS_Store or editor artifacts were added, and one or two recorded node refs still resolve if the sandbox may have changed. Dry-run every blueprint under references/blueprints/: mps_mcp_insert_root_node_from_json with dryRun: true into a sandbox model for a root blueprint, mps_mcp_update_node ADD CHILD with dryRun: true under a sample parent for a fragment. The server reads a json / childJson file only from the system temp directory, so a path under .agents/skills/ is rejected with "is not inside the system temp directory"; that rejection is not a blueprint failure. Send each file's exact content inline (up to 4,096 characters), or copy the files first and pass each copy's absolute path: d="${TMPDIR:-/tmp}/<dsl-name>-blueprints"; mkdir -p "$d"; cp .agents/skills/<dsl-name>-dsl/references/blueprints/*.json "$d/". When a dry run fails, fix the shipped file, not only the inline text or the temp copy, and dry-run again from the fixed file. Parsing as JSON is not enough, because the server rejects blueprints that parse. If every blueprint is rejected with Unknown property …, the language descriptor is hollow: MAKE the language and retry. Warnings for references given by name are expected; see "Dry-run response" in references/reference-formats/response-envelope.md in the mps-mcp-workflow skill root after loading that companion skill from the same origin..agents/skills/<dsl-name>-dsl/
|-- SKILL.md
`-- references/
|-- concepts.md
|-- sandbox.md
|-- workflows.md
|-- gotchas.md
`-- blueprints/
|-- <operation>-skeleton.json
`-- <operation>-subtree.jsonThe <dsl-name> slug is derived from the language namespace: use stable lowercase, split camel case and separators into words, and prefer the namespace's final meaningful segment unless the project consistently names the DSL differently. The -dsl suffix is mandatory — callers (such as mps-mcp-workflow) discover generated skills with the glob .agents/skills/*-dsl/.
SKILL.md Requirementsname: <dsl-name>-dsldescription: Use when creating, editing, validating, or inspecting this DSL's models..mps XML.mps_mcp_get_concept_details with l:<uuid>:<languageName>, not <uuid>(<languageName>).mps_mcp_check_root_node_problems.references/blueprints/ for known shapes: send a blueprint's content inline, or copy the file to ${TMPDIR:-/tmp} (%TEMP% on Windows) and pass the copy's absolute path. The server does not read a file under .agents/skills/.mps_mcp_insert_root_node_from_json.mps_mcp_insert_root_node_from_json.mps_mcp_check_root_node_problems and any task-required build/generation checks.concepts.md under the generated skill's references/: concept hierarchy, rootable concepts, properties, child roles, references, and semantic notes.sandbox.md under the generated skill's references/: sandbox model refs, representative roots, configuration nodes, reference targets, and stable node refs.workflows.md under the generated skill's references/: creation/editing recipes, including when to use full-root JSON versus skeleton-plus-subtrees.gotchas.md under the generated skill's references/: reference formats, ordering constraints, required roles, expression precedence issues, extension-language dependencies, and known validation failures.references/blueprints/: valid compact JSON skeletons and subtree templates, in the array form print_node emits (properties:[{name,value}], not a map).JetBrains/MPS
97d1e12
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.