@zenuml/core 3.47.9 → 3.48.0

package/.claude/skills/zenuml-ux-research/references/report-template.md

@@ -1,89 +0,0 @@
----
-scenario_id: {{scenario_id}}
-scenario_title: {{scenario_title}}
-run_date: {{run_date}}
-zenuml_core_sha: {{zenuml_core_sha}}
-audited_url: {{audited_url}}
-skill_version: 0.1.0
-gap_count: { high: {{gap_count_high}}, med: {{gap_count_med}}, low: {{gap_count_low}} }
----
-
-# UX Research — {{scenario_title}}
-
-## Executive summary
-
-{{executive_summary}}
-
-## Scenario recap
-
-**User intent:** {{user_intent}}
-
-**Starting DSL:**
-```
-{{starting_dsl}}
-```
-
-**Target DSL:**
-```
-{{target_dsl}}
-```
-
-## Observed walkthrough
-
-{{walkthrough_numbered_steps}}
-
-## Gaps
-
-{{gap_blocks}}
-
-> Each gap block uses this structure:
->
-> ```
-> ### Gap N — <short headline>
-> **Severity:** <low|med|high>
-> **Catalog ID:** <id or "novel — candidate for new rule">
-> **Observed:** <verbatim>
-> **Expected:** <from hypothesis>
-> **Exemplars:** <tools where expected behavior is seen>
-> **Rationale:** <why this matters>
-> **Suggested fix:** <grep result: file:line; or "no code path found">
-> ```
->
-> (Blockquote is for documentation inside the template — the rendered report omits it.)
-
-## Coverage
-
-Tested hypotheses (no gap found):
-{{coverage_tested}}
-
-Not tested (out of scope for this scenario):
-{{coverage_out_of_scope}}
-
-Skipped (couldn't form a testable hypothesis):
-{{coverage_skipped}}
-
-## Best-practice sources
-
-**Bundled catalog IDs referenced:**
-{{bundled_sources}}
-
-**Web sources fetched during this run:**
-{{web_sources}}
-
-## Playwright regression snippet
-
-Paste into `zenuml-core/tests/ux/{{scenario_id}}.spec.ts` once the gap is fixed. The skill emits this snippet; it does not run it.
-
-```typescript
-{{playwright_snippet}}
-```
-
----
-
-## Zero-gap rendering
-
-If the scenario produced zero gaps, render the report omitting the **Gaps** and **Playwright regression snippet** sections. The **Observed walkthrough** section becomes a single line:
-
-> No gaps observed on zenuml-core @ {{zenuml_core_sha}}.
-
-All other sections (metadata, executive summary, scenario recap, coverage, best-practice sources) still render.

package/.claude/skills/zenuml-ux-research/references/scenarios/edit-message-label.md

@@ -1,37 +0,0 @@
----
-id: edit-message-label
-title: Edit an existing message label
----
-
-## User intent
-The user has a message `A->B: hello` on the canvas and wants to change the label to `hi`.
-
-## Starting DSL
-```
-A
-B
-A->B: hello
-```
-
-## Target DSL
-```
-A
-B
-A->B: hi
-```
-
-## Relevant assertion categories
-EDT, KBD, SEL, FOC, UND
-
-## Walkthrough hints (not prescriptive)
-- This scenario exercises the inline-edit contract on an existing piece of DSL, not a new one.
-- Candidate paths to try in order:
-  1. Click the message label `hello` once to select, press Enter, retype.
-  2. Double-click the message label.
-  3. Keyboard: navigate to the message via arrows, press Enter.
-  4. Fall back to editing the DSL editor directly.
-- Watch specifically for: undo granularity when the user types `hi` then changes their mind (one undo should revert the whole label, not one character at a time).
-
-## Known issues to watch for (optional)
-- If undo is character-level during a label edit, that is a UND-02 violation.
-- If clicking the message selects the wrong thing (e.g., the arrow instead of the label), that is a SEL-01 violation.

package/.claude/skills/zenuml-ux-research/references/scenarios/insert-message.md

@@ -1,36 +0,0 @@
----
-id: insert-message
-title: Insert a synchronous message between two participants
----
-
-## User intent
-The user has two participants `A` and `B` on the canvas and wants to add a synchronous message from `A` to `B` with the label `hello`.
-
-## Starting DSL
-```
-A
-B
-```
-
-## Target DSL
-```
-A
-B
-A->B: hello
-```
-
-## Relevant assertion categories
-INS, KBD, EDT, FOC, FBK
-
-## Walkthrough hints (not prescriptive)
-- This is the most common creation action in any sequence diagramming tool; it is the load-bearing scenario for canvas-first editing of ZenUML.
-- Candidate paths to try in order:
-  1. Click `A`, look for a "send message" affordance or toolbar button, draw/click toward `B`.
-  2. Hover between `A` and `B` in the lifeline area for an inline "+" affordance.
-  3. Keyboard: Tab into widget, select `A`, press a hotkey for "new message".
-  4. Fall back to typing `A->B: hello` into the DSL editor.
-- Watch specifically for: whether a canvas-first path exists at all, whether message creation automatically enters label edit mode on the new message, whether Escape during message creation cancels cleanly.
-
-## Known issues to watch for (optional)
-- If the canvas has no message-creation affordance at all, that is an INS-01 violation at high severity.
-- If the new message is not in edit mode immediately on creation, that is an EDT-03 / FOC-01 violation.

package/.claude/skills/zenuml-ux-research/references/scenarios/insert-participant.md

@@ -1,31 +0,0 @@
----
-id: insert-participant
-title: Insert a participant on a blank diagram
----
-
-## User intent
-The user opens ZenUML to a blank diagram and wants to add one participant named `Alice` so they can start modelling.
-
-## Starting DSL
-```
-```
-
-## Target DSL
-```
-Alice
-```
-
-## Relevant assertion categories
-INS, FOC, EDT, KBD
-
-## Walkthrough hints (not prescriptive)
-- Blank canvas is the highest-stakes discoverability test — the skill should record how long it takes to find the insertion affordance.
-- Candidate insertion paths to try in order:
-  1. Click on the canvas background.
-  2. Look for a visible "+" or "Add participant" affordance.
-  3. Keyboard: try pressing Enter or `p` on an empty canvas.
-  4. Fall back to typing directly into the DSL editor pane.
-- The moment the user successfully names the new participant `Alice` is the scenario's end state.
-
-## Known issues to watch for (optional)
-- If the only path is direct DSL editing, that itself is a finding (violates INS-01).

package/.claude/skills/zenuml-ux-research/references/scenarios/rename-participant.md

@@ -1,33 +0,0 @@
----
-id: rename-participant
-title: Rename a participant via keyboard
----
-
-## User intent
-The user has a participant `A` on the canvas and wants to rename it to `Alice` without leaving the keyboard.
-
-## Starting DSL
-```
-A
-```
-
-## Target DSL
-```
-Alice
-```
-
-## Relevant assertion categories
-KBD, EDT, SEL, FOC
-
-## Walkthrough hints (not prescriptive)
-- This scenario is the canonical test of KBD-03 (Enter enters edit mode) and EDT-02 (caret at end of text).
-- Candidate paths to try in order:
-  1. Tab into the diagram widget, arrow-key to select `A`, press Enter, type `lice`, press Enter.
-  2. Click `A` to select, press Enter, retype.
-  3. Double-click `A` to enter edit mode.
-  4. Fall back to editing the DSL editor directly.
-- Watch specifically for: whether Enter does anything on the selected participant; whether the caret is placed at the end or the whole label is pre-selected; whether Escape cancels cleanly.
-
-## Known issues to watch for (optional)
-- If Enter does nothing on a selected participant, that is a KBD-03 violation at high severity.
-- If only mouse double-click works, that is a KBD-only violation (no mouse-free path).

package/.claude/skills/zenuml-ux-research/references/scenarios/undo-insert.md

@@ -1,35 +0,0 @@
----
-id: undo-insert
-title: Undo a just-inserted message
----
-
-## User intent
-The user has two participants `A` and `B`, inserts a message `A->B: hello`, then immediately presses Ctrl/Cmd+Z to undo. The expected result is that the message is removed and the state is back to just the two participants.
-
-## Starting DSL
-```
-A
-B
-```
-
-## Target DSL
-```
-A
-B
-```
-(after: insert `A->B: hello`, then undo)
-
-## Relevant assertion categories
-UND, FOC, SEL, KBD
-
-## Walkthrough hints (not prescriptive)
-- This scenario exercises the full insertion-then-undo round trip.
-- Steps:
-  1. Follow the `insert-message` scenario's walkthrough to get to the post-insert state.
-  2. Press Ctrl+Z (or Cmd+Z on macOS).
-  3. Observe: is the message removed? Is the DSL editor reverted? Is selection/focus restored to whatever it was before the insert?
-- Watch specifically for: undo granularity (one Ctrl+Z should undo the whole insert, not the individual keystrokes of the label edit) and focus restoration (FOC-03).
-
-## Known issues to watch for (optional)
-- If the undo leaves the label partially typed, UND-02 is violated.
-- If focus ends up on the document body after undo, FOC-03 is violated.

package/.devcontainer/devcontainer.json

@@ -1,21 +0,0 @@
-{
-  "name": "Node.js & Browser Tools",
-  "image": "mcr.microsoft.com/devcontainers/typescript-node:20",
-  "features": {
-    "ghcr.io/devcontainers/features/node:2": {
-      "version": "none"
-    }
-  },
-  "forwardPorts": [9323],
-  "customizations": {
-    "vscode": {
-      "extensions": [
-        "dbaeumer.vscode-eslint",
-        "esbenp.prettier-vscode"
-      ]
-    }
-  },
-  // Add your source code to the container
-  "postCreateCommand": "pnpm install",
-  "remoteUser": "node"
-}

package/.dockerignore

@@ -1,19 +0,0 @@
-node_modules
-.git
-.gitignore
-.github
-*.md
-dist
-.watchmanconfig
-.prettierrc
-.prettierignore
-.editorconfig
-.eslintignore
-.eslintrc.js
-.git-blame-ignore-revs
-.vscode
-cypress
-cypress.config.ts
-docs
-test
-vite.config.lib.js

package/.eslintrc.js

@@ -1,39 +0,0 @@
-module.exports = {
-  env: {
-    browser: true,
-    es2021: true,
-    node: true,
-    jest: true,
-  },
-  extends: [
-    "eslint:recommended",
-    "plugin:@typescript-eslint/recommended",
-    "plugin:prettier/recommended",
-  ],
-  globals: {
-    CodeMirror: "readonly",
-  },
-  overrides: [
-    {
-      env: {
-        node: true,
-      },
-      files: [".eslintrc.{js,cjs}"],
-      parserOptions: {
-        sourceType: "script",
-      },
-    },
-  ],
-  parserOptions: {
-    ecmaVersion: "latest",
-    parser: "@typescript-eslint/parser",
-    sourceType: "module",
-  },
-  plugins: ["@typescript-eslint", "html"],
-  rules: {
-    "@typescript-eslint/ban-ts-comment": "off",
-    "@typescript-eslint/no-var-requires": "off",
-    "@typescript-eslint/no-explicit-any": "off",
-    "@typescript-eslint/no-this-alias": "off",
-  },
-};

package/.git-blame-ignore-revs

@@ -1,6 +0,0 @@
-# This file is used to configure to ignore the given commit sha in git blame
-# To apply this file, run the following command:
-# git config blame.ignorerevsfile .git-blame-ignore-revs
-
-# This commit formatted the entire codebase
-80987154e7740ddae2c7c0a04607bc25da51586e

package/.kiro/hooks/README.md

@@ -1,38 +0,0 @@
-# Kiro Agent Hooks
-
-This directory contains custom hooks that automatically execute when specific events occur in Kiro.
-
-## Session Sound Notification Hook
-
-**File:** `session-sound-notification.json` and `session-sound-notification.js`
-
-**Purpose:** Plays system sounds to notify you when:
-
-- An AI session completes
-- User input is required to continue
-- An error occurs
-
-**Sounds Used (macOS):**
-
-- **Session Complete:** Glass.aiff - A pleasant chime when tasks finish
-- **Input Required:** Sosumi.aiff - An attention-getting sound when you need to respond
-- **Error:** Basso.aiff - A distinctive sound for errors
-- **Default:** Ping.aiff - Fallback sound
-
-**Configuration:**
-The hook is enabled by default and will auto-execute. You can disable it by setting `"enabled": false` in the JSON configuration file.
-
-**Platform Notes:**
-
-- Currently optimized for macOS using the `afplay` command
-- Uses built-in system sounds located in `/System/Library/Sounds/`
-- For other platforms, the sound files and playback command would need to be adjusted
-
-## Managing Hooks
-
-You can:
-
-1. View current hooks in the Kiro Explorer under "Agent Hooks"
-2. Use Command Palette → "Open Kiro Hook UI" to create new hooks
-3. Edit hook files directly in this directory
-4. Enable/disable hooks by modifying the `enabled` property

package/.kiro/hooks/session-sound-notification.js

@@ -1,44 +0,0 @@
-/**
- * Session Sound Notification Hook
- * Plays a sound when sessions end or user input is required
- */
-
-const playSound = (soundType = 'default') => {
-  // For macOS, we can use the 'afplay' command to play system sounds
-  const sounds = {
-    default: '/System/Library/Sounds/Ping.aiff',
-    complete: '/System/Library/Sounds/Glass.aiff',
-    attention: '/System/Library/Sounds/Sosumi.aiff',
-    error: '/System/Library/Sounds/Basso.aiff'
-  };
-  
-  const soundFile = sounds[soundType] || sounds.default;
-  
-  try {
-    // Use afplay to play the sound on macOS
-    require('child_process').exec(`afplay "${soundFile}"`, (error) => {
-      if (error) {
-        console.log('Could not play sound:', error.message);
-      }
-    });
-  } catch (err) {
-    console.log('Sound playback failed:', err.message);
-  }
-};
-
-module.exports = {
-  onSessionEnd: () => {
-    console.log('🔊 Session completed - playing completion sound');
-    playSound('complete');
-  },
-  
-  onUserInputRequired: () => {
-    console.log('🔊 User input needed - playing attention sound');
-    playSound('attention');
-  },
-  
-  onError: () => {
-    console.log('🔊 Error occurred - playing error sound');
-    playSound('error');
-  }
-};

package/.kiro/hooks/session-sound-notification.json

@@ -1,23 +0,0 @@
-{
-  "name": "Session Sound Notification",
-  "description": "Plays a sound when a session ends or user input is needed",
-  "triggers": [
-    {
-      "event": "session.end",
-      "description": "Triggered when an AI session completes"
-    },
-    {
-      "event": "user.input.required",
-      "description": "Triggered when the AI needs user input to continue"
-    }
-  ],
-  "actions": [
-    {
-      "type": "system.sound",
-      "sound": "notification",
-      "description": "Play system notification sound"
-    }
-  ],
-  "enabled": true,
-  "autoExecute": true
-}

package/.mcp.json.example

@@ -1,17 +0,0 @@
-{
-  "mcpServers": {
-    "playwright": {
-      "command": "npx",
-      "args": ["@playwright/mcp@latest", "--extension"],
-      "env": {
-        "PLAYWRIGHT_MCP_EXTENSION_TOKEN": "<your-token-here>"
-      }
-    },
-    "chrome-devtools": {
-      "type": "stdio",
-      "command": "npx",
-      "args": ["-y", "chrome-devtools-mcp"],
-      "env": {}
-    }
-  }
-}

package/.nvmrc

@@ -1 +0,0 @@
-20

package/.prettierignore

@@ -1,4 +0,0 @@
-node_modules
-dist
-generated-parser
-pnpm-lcok.yaml

package/.prettierrc

@@ -1 +0,0 @@
-{}

package/.specify/memory/constitution.md

@@ -1,33 +0,0 @@
-<!-- Sync Impact Report
-Version change: 1.0.0 → 1.1.0 (New principle added)
-Modified principles: None
-Added sections: II. Task Focus principle
-Removed sections: None
-Templates requiring updates: ✅ plan-template.md
-Follow-up TODOs: RATIFICATION_DATE needs confirmation
--->
-
-# ZenUML Core Constitution
-
-## Core Principles
-
-### I. Parser Simplicity
-
-The ANTLR lexer and parser MUST remain simple and maintainable. Edge cases and
-complex scenarios SHALL be handled in utility methods or the renderer, not in the
-grammar. Reasonable fallbacks are preferred over grammar complexity. This ensures
-the parser remains understandable, testable, and maintainable while avoiding
-fragility from over-engineering.
-
-### II. Task Focus
-
-When working on one task, focus on that task. Do not try to fix unrelated issues
-(code smell, inconsistency, or even defects). You SHALL mention unrelated issues
-but MUST NOT fix them. This ensures changes remain scoped, reviewable, and
-traceable to specific requirements.
-
-## Governance
-
-The Constitution supersedes all other development practices. Amendments require documentation and versioning.
-
-**Version**: 1.1.0 | **Ratified**: TODO(RATIFICATION_DATE): Needs confirmation | **Last Amended**: 2025-10-06