Miragon

Write or refactor unit tests following the canonical project pattern. Use when adding new unit tests or standardizing existing ones.

Generate a Kotlin SingleModelValidationRule implementation for ONE rule from BPMN_STYLE_GUIDE.md. Use when drafting / iterating on a single rule, or when the user asks to 'generate a rule for <slug>', 'create the element-id-format rule', or 'add just this one rule to the test module'. For all rules at once, use generate-rules-to-enforce-bpmn-styleguide.

Generate Kotlin SingleModelValidationRule implementations for every deterministic (or hybrid) rule in BPMN_STYLE_GUIDE.md, so they can be enforced at test time by the bpmn-to-code-testing module. llm-only rules stay in the style guide for /validate-bpmn-style. Use when the user asks to 'generate validation rules from the style guide', 'enforce BPMN conventions in CI', or 'automate style checks in tests'. For a single rule, use generate-rule-to-enforce-bpmn-styleguide.

Interactively author a BPMN_STYLE_GUIDE.md for your project — a handbook that explains how the team models BPMN processes (naming, allowed elements, IDs, topics, engine-specific special cases). The output is a document humans can read; embedded YAML annotations let downstream skills enforce the rules. Use when the user asks to 'create a BPMN style guide', 'set up BPMN conventions', or 'define BPMN naming rules'.

Create a complete Spring Boot service project from a BPMN file with hexagonal architecture. Bootstraps via Spring Initializr, configures bpmn-to-code, and generates workers, use cases, and services wired to ProcessApi constants — no raw strings anywhere. Use when the user wants to start a new process service from a BPMN model.

Migrate generated API usage from bpmn-to-code v1.1.0 to v2.0.0. Use when the user asks to 'upgrade to v2', 'migrate from v1 to v2', 'fix TaskTypes references', or 'update bpmn-to-code API calls after upgrade'.

Replace hardcoded BPMN strings with references to the generated Process API. Use when the user asks to 'migrate to the process API', 'replace hardcoded strings with API references', 'use the generated BPMN API', or 'refactor to type-safe BPMN constants'.

Set up the bpmn-to-code Gradle plugin in an existing project. Adds the plugin dependency and configures the code generation task. Use when the user asks to 'add bpmn-to-code to my Gradle project', 'set up BPMN code generation', or 'configure the Gradle plugin'.

Internal architecture of the modeler engine and its hosts — the host-agnostic `libs/modeler-core` package (domain/service/vscode-free infrastructure) consumed by the `apps/vscode-plugin` extension host, the IntelliJ plugin (via `apps/modeler-bridge`), and the Theia/Electron standalone. Covers the feature-folder + four-layer split (domain, infrastructure, service, controller), host-capability ports + hexagonal engine ports, deployment subsystem (Camunda 7 & 8), per-feature register() composition root, WebviewMessageRouter dispatch, EditorSessionParticipant lifecycle, EditorSessionStore + VsCodeEditorHandle, echo-prevention session guards, codeLink + template-marketplace features, webview message protocol (Query/Command), and the ArchUnitTS architecture tests. Use this skill when working on extension-host or modeler-core code — implementing features, fixing bugs, reviewing PRs, refactoring services, adding message types, understanding editor tracking, tracing webview message flow, integrating external systems,

Interact with the BPMN modeler webview running in a browser using Playwright MCP tools. Use this skill whenever the user asks to test, inspect, or interact with the BPMN modeler in a browser — including adding/modifying BPMN elements, checking the properties panel, verifying UI behavior, or debugging the webview. Also trigger when the user mentions "open the modeler in a browser", "add a task/event/gateway", "check the properties panel", or any visual/interactive testing of the BPMN webview.

The IntelliJ host for the BPMN modeler (apps/intellij-plugin, Kotlin/Gradle) and the apps/modeler-bridge stdio JSON-RPC Bun binary that runs the unmodified @miragon/bpmn-modeler-core out-of-process. Covers the Gradle build/run order (build:libs → build:bpmn-webview → bridge compile → ./gradlew runIde), webview bundle staging and the blank-webview trap, the version-keyed bridge-binary cache trap and the -Dmiragon.bridge dev override, the JCEF editor + JBCefJSQuery message relay, the RPC method contract and host-replicated mirrors (document/settings/deployment), plugin.xml extension points, and the process supervisor. Use this skill whenever working on the IntelliJ plugin, the modeler-bridge, JCEF editors/tool windows, plugin.xml, Gradle tasks, RPC adapters, a blank IntelliJ webview, verifyPlugin, or debugging the IntelliJ↔bridge seam. See the architecture skill for the host-agnostic core these two consume.

VS Code CustomTextEditorProvider pattern — registration, document sync, lifecycle, disposables, the generic ModelerEditorController + EditorSessionParticipant pattern, BPMN/DMN differences. Use this skill whenever working on the editor controller, document synchronization, CustomTextEditorProvider, webview lifecycle, editor registration, resolveCustomTextEditor, EditorSessionStore, session participants, or understanding how BPMN/DMN files are opened and edited. Also consult this skill when debugging issues with editor state, hidden webviews, tab switching, or disposable cleanup — even if the user doesn't mention "custom editors" explicitly.

VS Code UX guidelines for this project — how to choose between notifications, status bar, quick picks, and output channels; how clipboard access works between sandboxed webviews and the extension host; webview theming; accessibility; and setContext for keybinding visibility. Use this skill whenever you are adding user-facing feedback (messages, errors, progress), wiring clipboard in webviews, implementing copy/paste, choosing how to surface information to the user, adding keyboard shortcuts or when-clause visibility, working on theming, or considering accessibility. Also use when deciding between notification types or between notifications vs. status bar vs. output channel.

VS Code webview internals — HTML serving, CSP, postMessage protocol, state persistence, theming. Use when working on webview HTML, postMessage, CSP, theme handling, vscode.getState/setState, acquireVsCodeApi, webview.asWebviewUri, retainContextWhenHidden, nonce generation, content security policy, or the Query/Command message protocol between extension host and webview.

bpmn-js modeler internals — EventBus, services, copy-paste architecture, clipboard polyfill, modeler lifecycle. Use when working on bpmn-webview, diagram interactions, copy-paste, clipboard, or element templates.

Add or complete translations for the BPMN/DMN modeler UI in the miranum-ide project. Use this skill whenever the user wants to add a new language, translate missing keys for an existing locale, update translations, or register a new locale in the i18n system. Also trigger when the user mentions locale codes, language names, or talks about translating modeler UI strings in this repository.

Conversational process capture — interviews a process owner who cannot write BPMN, elicits trigger, outcome, happy path, decisions, handoffs, and exceptions one question at a time, plays the understood process back in plain language for correction, then hands the confirmed data to the new-process skill for scaffolding. Use when someone wants to talk through, describe, or explain how their process works, wants a process documented from an interview, says "let me tell you how we do X", or says they don't know BPMN.

Exports a business process — its BPMN model, any sub-processes it calls, and the derived process view — as a self-contained, portable agent skill under dist/skills/. Use when the user wants a process "as a skill", wants processes to be usable by other agents or projects, or says "let this process talk".

Imports existing process documentation — draw.io or Visio XML, pasted Word/Confluence text, flowchart images the agent reads, PowerPoint exports — into the repository as a BPMN model. Extracts activities, decisions, and roles; generates a convention-conform .bpmn with a complete diagram layout; records provenance and runs the validator. Use when the user wants to import, convert, migrate, or digitize an existing process document, diagram, or wiki page.

Scaffolds a new business process in this repository — creates a single BPMN file under the processes folder with a complete diagram (semantics + layout), lanes for the owning roles, and callActivity links to any sub-processes. Use when the process facts (trigger, outcome, roles, rough flow) are already known — from the user or a capture-process interview — and the user wants to add, scaffold, or start modeling a new process.

Files and triages reports that a process model no longer matches reality — the inbound channel of "let your processes talk". Use when someone says "that's not how we do it anymore", "this step is outdated", or "we skip that check now" (in any conversation, about any process), when the user pastes a PROCESS FEEDBACK block produced by an exported process skill, or when they ask to file, list, triage, accept, or reject process feedback. Owns the feedback/<id>/ inbox convention and drives accepted corrections through a model edit.

Lets the user talk to the processes in this repository. Answers questions about modeled business processes — how a process works step by step, who does what (from the BPMN lanes), what it calls, and what is affected by a change. Use when the user asks about a process, a role/ownership question, an impact analysis, or wants a walkthrough.

Quality gate for process models — runs the deterministic validator (BPMN structure + BPMNDI coverage + callActivity link integrity), then checks BPMN modeling conventions and diagram clarity by judgment. Use when the user asks to review, validate, lint, or check a process, before merging model changes, or for a portfolio health report.

Step-by-step house pattern for adding a new analytics capability — a PromQL query function in `packages/client-analytics`, a tool in `packages/mcp-analytics`, and optionally a dashboard widget. Use whenever adding or changing process analytics ("add a query for X", "new KPI/metric analysis", "extend the analytics dashboard"), PromQL queries, Prometheus-backed tools, or analytics widgets. Takes precedence over the generic mcp-apps-builder skill.

Step-by-step house pattern for adding a new BPM operations tool, widget tool, or widget to the camunda7 module (`packages/mcp-camunda7` + `packages/client-camunda7`). Use whenever adding or changing Camunda 7 / CIB Seven tools ("add a tool to suspend jobs", "expose X from the engine REST API"), input schemas, `show_*` widget tools, `*_data` feeds, or React widgets in the cockpit. Takes precedence over the generic mcp-apps-builder skill.

Always-on style guide for the Miragon AI Platform documentation under `docs/`. Use whenever editing, reviewing, or planning anything in `docs/` (any `.md` file, `.vitepress/config.ts`, `package.json` of `@miragon-ai/docs`), whenever shipping a feature that changes user-visible behavior (env vars, modules, tools, deployment), and proactively whenever you suspect the docs site has drifted from the code. Encodes the four-section structure (Architecture / Developer / Operations / Usage), per-page line budgets, English-only rule, the VitePress + Mermaid stack, and the rule against duplicating the root `README.md`.

**MANDATORY for ALL MCP server work** - mcp-use framework best practices and patterns. **READ THIS FIRST** before any MCP server work, including: - Creating new MCP servers - Modifying existing MCP servers (adding/updating tools, resources, prompts, widgets) - Debugging MCP server issues or errors - Reviewing MCP server code for quality, security, or performance - Answering questions about MCP development or mcp-use patterns - Making ANY changes to server.tool(), server.resource(), server.prompt(), or widgets This skill contains critical architecture decisions, security patterns, and common pitfalls. Always consult the relevant reference files BEFORE implementing MCP features.

Create or update a GitHub issue for automation-mcp using the `gh` CLI. Use when the user asks to "file a bug", "create a feature request", "open a GitHub issue", or "update an existing one". Supports feature, bug, and refactor issue types with structured templates; shows a draft for confirmation before creating or editing; looks up existing issues by number or URL for updates.

Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI. Provides project context, component docs, and usage examples. Applies when working with shadcn/ui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for "shadcn init", "create an app with --preset", or "switch to --preset".

Build an MCP server with @miragon/mcp-toolkit: a host via createFrameworkApp plus a module that registers its OWN tools and a widget. Use when asked to "build/create an MCP server", "stand up an MCP server with tools and a UI", "add a module with its own tools" (createToolRegistrar), "expose a tool plus a widget", wire an AppPlugin into createFrameworkApp, add an app-only *_data feed (APP_ONLY_META), write a show_* widget tool (buildSingleWidgetView / uiMeta), or bundle a widget into mcp-app.html. The worked example is the `tasks` module.

Build an MCP widget against @miragon/mcp-toolkit-ui. Use when asked to build a widget, render a tool result as UI, add a card/panel/dashboard for an MCP tool, write a *_show_* widget, or turn structuredContent into a React component in this repo. Encodes the data contract (tool result -> ViewStructuredContent -> data prop), the component catalog, host-portable authoring (useHostBridge), and the widget-playground iterate loop.

Compose a multi-widget view or build a pipeline with @miragon/mcp-toolkit. Use when asked to build a dashboard, compose/lay out several widgets in one view, write a multi-widget *_show_* tool, chain tool outputs through a multi-step pipeline, drive render-view, define a layout (rows/cells/span) or per-cell props, or decide between an eager view and a pipeline. The worked example is the `orders` module. Encodes the eager buildComposedView default, the advanced render-view pipeline, the requires/_dataType double contract, and the layout format.

Add a tool to an existing @miragon/mcp-toolkit MCP module. Use when asked to "add a tool", "expose X as a tool", "register a new MCP tool", write a new createToolRegistrar entry, set tool annotations (readOnly/idempotent/destructive hints), add an outputSchema, paginate a list tool, or add an app-only *_data feed (APP_ONLY_META) for a widget. For a whole new server/module use build-mcp-server; for the UI use build-mcp-widget.

White-label a client UI with the @miragon/mcp-toolkit theming system. Use when asked to "re-brand"/"white-label" a UI, apply a client's brand colours, set up createTheme / ThemeProvider / themePresets, support light/dark mode, theme a widget or app for a specific customer, or fix hard-coded colours that ignore the brand. The rule: use theme tokens (text-primary, bg-card, rounded-lg), never hard-coded colours.

Synchronisiert Aufträge aus Dimacon nach Clockin als Projekte. Verwende diesen Skill immer wenn der Benutzer Dimacon-Aufträge nach Clockin übertragen, Clockin-Projekte aus Dimacon erstellen, die Tagesplanung synchronisieren, oder Aufträge in Clockin anlegen möchte. Auch bei Formulierungen wie "Clockin sync", "Aufträge nach Clockin", "Clockin aktualisieren", "Projekte für heute anlegen", "sync die Aufträge", "Clockin CSV", "Tagesplanung übertragen" oder "was steht heute an und trag es in Clockin ein" soll dieser Skill verwendet werden. Der Skill holt alle Termine für ein Datum aus Dimacon, prüft ob die Projekte in Clockin bereits existieren, und erstellt oder aktualisiert sie mit korrekten Mitarbeiter-Zuweisungen.

Plant neue Aufträge in Dimacon ein – vom Kundenabgleich über Projektauswahl bis zur Job- und Working-Erstellung. Verwende diesen Skill IMMER wenn der Benutzer einen Auftrag einplanen, anlegen, erstellen oder terminieren will. Auch verwenden bei: "Plane einen Auftrag für [Kunde]", "Leg einen Job an", "Neuer Auftrag für [Projekt]", "Termin einplanen", "Auftrag erstellen", "Job anlegen für [Datum]", "Kernbohrung einplanen", "Wandsäge-Auftrag", "Plane das ein", "Kannst du das einplanen?", "Auftrag aus Angebot erstellen", "Mach einen Job draus", "Plane [Angebotsnummer] ein". Triggert auch wenn der Benutzer eine Lexware-Angebotsnummer (z.B. "AG-2026-042") nennt und einen Auftrag daraus machen will, oder wenn er einen Kunden + Datum nennt und etwas eingeplant haben möchte. NICHT verwenden für Abrechnung (→ dimacon:abrechnung) oder Angebotserstellung aus Planungsexport-URLs (→ dimacon:lexware).

Erstellt Angebote in Lexware Office (Lexoffice) aus einem hochgeladenen LV (PDF). Daraus wird eine strukturierte Excel (Claude Desktop/Code) bzw. ein editierbares Live-Artefakt (Cowork), das per Knopfdruck zum Lexoffice-Angebot wird. Triggert bei "hier ist das LV", "kalkulier mir das LV", "Excel zum LV", "Angebot aus LV", oder wenn ein PDF mit "LV", "Leistungsverzeichnis" oder "Ausschreibung" hochgeladen wird. NICHT für: einfache Angebote ohne LV (→ lexoffice:angebot), Rechnungen aus Dimacon-Jobs (→ dimacon:abrechnung), Angebote aus Dimacon-Planungs-URLs (→ dimacon:lexware), Aufträge planen (→ dimacon:planung).

Erstellt Angebote in Lexware Office (Lexoffice) mit korrekter Kundenanlage, Steuerlogik (§13b für Bauleistende) und Deeplink-Ausgabe. Diesen Skill IMMER verwenden, wenn der User ein Angebot, Kostenvoranschlag oder Quotation in Lexware/Lexoffice erstellen will — auch wenn er nur sagt "leg ein Angebot an", "mach mir ein Angebot für [Kunde]", "Lexware Angebot", "neuer Auftrag in Lexoffice", "Kostenvoranschlag schreiben", "Bauleistung anbieten", "Offer schreiben". Triggert auch, wenn er einen neuen Kunden + Positionen + Bauvorhaben in einem Schritt nennt. NICHT verwenden für: Rechnungen aus Dimacon-Jobs (→ dimacon:abrechnung), Angebote aus Dimacon-Planungs-URLs (→ dimacon:lexware), neue Aufträge planen (→ dimacon:planung), LV-basierte Angebote (→ lexoffice:angebot-lv).

**MANDATORY for ALL MCP server work** - mcp-use framework best practices and patterns. **READ THIS FIRST** before any MCP server work, including: - Creating new MCP servers - Modifying existing MCP servers (adding/updating tools, resources, prompts, widgets) - Debugging MCP server issues or errors - Reviewing MCP server code for quality, security, or performance - Answering questions about MCP development or mcp-use patterns - Making ANY changes to server.tool(), server.resource(), server.prompt(), or widgets This skill contains critical architecture decisions, security patterns, and common pitfalls. Always consult the relevant reference files BEFORE implementing MCP features.

DEPRECATED: This skill has been replaced by `mcp-app-builder`. Check if `mcp-app-builder` is available in the skills folder. If not, install it: `npx skills install mcp-use/mcp-use --skill mcp-app-builder` Use `mcp-app-builder` instead of this skill. Build Model Context Protocol (MCP) servers with mcp-use framework. Use when creating MCP servers, defining tools/resources/prompts, working with mcp-use, bootstrapping MCP projects, or when user mentions MCP development, tools, resources, or prompts.

Translate Camunda Modeler i18n plugin keys into a target language, using the German translation as the authoritative base. Use this skill whenever the user wants to add a new language, translate keys, fill in missing translations for an existing language, or update/complete a partial translation in the camunda-modeler-i18n-plugin repository. Also trigger when the user mentions locale codes, language names, or talks about translating the modeler UI.

Wardley Mapping expert. Explains Wardley Maps and the Online-Wardley-Maps (OWM) text DSL, teaches Doctrine (universal principles) and leadership/gameplay (context-specific plays), and reviews a user's map — what is good, what is weak, and where the potential is — adding colour-coded feedback notes directly to the .wmap/.owm file. Use whenever the user asks to review, critique, improve, or sanity-check a Wardley map; to explain Wardley-mapping concepts (evolution, doctrine, climatic patterns, pioneers/settlers/town planners, the strategy cycle); or to read/write .wmap / .owm files.

Value chain diagram (Wertschöpfungskettendiagramm, ARIS VACD) expert. Explains the notation (chevron steps, sequence/hierarchy relations, organizational units), reads and writes the project's *.vc.json document format, and reviews a user's process landscape — structure, granularity, decomposition depth, and responsibility assignment. Use whenever the user asks to create, review, critique, or restructure a value chain diagram; to explain the WKD/VACD notation; or to read/write *.vc.json files.