moai-adk 0.4.5__py3-none-any.whl → 0.4.8__py3-none-any.whl

moai_adk/templates/.claude/skills/moai-alfred-refactoring-coach/SKILL.md

@@ -1,97 +0,0 @@
----
-
-name: moai-alfred-refactoring-coach
-description: Refactoring guidance with design patterns, code smells detection, and step-by-step improvement plans. Use when outlining refactor steps that preserve functionality.
-allowed-tools:
-  - Read
-  - Write
-  - Edit
-  - Bash
-  - TodoWrite
----
-
-# Alfred Refactoring Coach
-
-## Skill Metadata
-| Field | Value |
-| ----- | ----- |
-| Allowed tools | Read (read_file), Write (write_file), Edit (edit_file), Bash (terminal), TodoWrite (todo_write) |
-| Auto-load | /alfred:2-run refactor lane |
-| Trigger cues | Refactoring retros, duplication cleanup, code health follow-ups inside Alfred. |
-
-## What it does
-
-Refactoring guidance with design pattern recommendations, code smell detection, and step-by-step improvement plans.
-
-## When to use
-
-- Activates when Alfred is asked to plan or stage refactoring work.
-- “Help with refactoring”, “How can I improve this code?”, “Apply design patterns” 
-- “Code organization”, “Remove duplication”, “Separate functions”
-
-## How it works
-
-**Refactoring Techniques**:
-- **Extract Method**: Separate long methods
-- **Replace Conditional with Polymorphism**: Remove conditional statements
-- **Introduce Parameter Object**: Group parameters
-- **Extract Class**: Massive class separation
-
-**Design Pattern Recommendations**:
-- Complex object creation → **Builder Pattern**
-- Type-specific behavior → **Strategy Pattern**
-- Global state → **Singleton Pattern**
-- Incompatible interfaces → **Adapter Pattern**
-- Delayed object creation → **Factory Pattern**
-
-**3-Strike Rule**:
-```
-1st occurrence: Just implement
-2nd occurrence: Notice similarity (leave as-is)
-3rd occurrence: Pattern confirmed → Refactor! 🔧
-```
-
-**Refactoring Checklist**:
-- [ ] All tests passing before refactoring
-- [ ] Code smells identified
-- [ ] Refactoring goal clear
-- [ ] Change one thing at a time
-- [ ] Run tests after each change
-- [ ] Commit frequently
-
-## Best Practices
-- The text shown to the user is written using TUI/report expressions.
-- When running the tool, a summary of commands and results are recorded.
-
-## Examples
-```markdown
-- Call this skill inside the /alfred command to generate a report.
-- Add summary to Completion Report.
-```
-
-## Inputs
-- MoAI-ADK project context (`.moai/project/`, `.claude/` templates, etc.).
-- Parameters passed from user commands or higher commands.
-
-## Outputs
-- Reports, checklists or recommendations for your Alfred workflow.
-- Structured data for subsequent subagent calls.
-
-## Failure Modes
-- When required input documents are missing or permissions are limited.
-- When disruptive changes are required without user approval.
-
-## Dependencies
-- Cooperation with higher-level agents such as cc-manager and project-manager is required.
-
-## References
-- Fowler, Martin. "Refactoring: Improving the Design of Existing Code." Addison-Wesley, 2018.
-- IEEE Software. "Managing Technical Debt." IEEE Software, 2021.
-
-## Changelog
-- 2025-03-29: Added input/output/failure response to Alfred-specific skills.
-
-## Works well with
-
-- alfred-code-reviewer
-- alfred-trust-validation

moai_adk/templates/.claude/skills/moai-alfred-tui-survey/SKILL.md

@@ -1,87 +0,0 @@
----
-
-name: moai-alfred-tui-survey
-description: Standardizes Claude Code Tools AskUserQuestion TUI menus for surveys, branching approvals, and option picking across Alfred workflows. Use when gathering approvals or decisions via Alfred’s TUI menus.
-allowed-tools:
-  - Read
-  - Write
-  - Edit
-  - TodoWrite
----
-
-# Alfred TUI Survey Skill
-
-## Skill Metadata
-| Field | Value |
-| ----- | ----- |
-| Allowed tools | Read (read_file), Write (write_file), Edit (edit_file), TodoWrite (todo_write) |
-| Auto-load | On demand when AskUserQuestion menus are built |
-| Trigger cues | Branch approvals, survey menus, decision gating via AskUserQuestion. |
-
-## What it does
-
-Provides ready-to-use patterns for Claude Code's AskUserQuestion TUI selector so Alfred agents can gather user choices, approvals, or survey answers with structured menus instead of ad-hoc text prompts.
-
-## When to use
-
-- Activates when Alfred needs to gather structured choices through the TUI selector.
-- Need confirmation before advancing to a risky/destructive step.
-- Choosing between alternative implementation paths or automation levels.
-- Collecting survey-like answers (persona, tech stack, priority, risk level).
-- Any time a branched workflow depends on user-selected options rather than free-form text.
-
-## How it works
-
-1. **Detect gate** – Pause at steps that require explicit user choice.
-2. **Shape options** – Offer 2–5 focused choices with concise labels.
-3. **Render menu** – Emit the `AskUserQuestion({...})` block for the selector.
-4. **Map follow-ups** – Note how each option alters the next action/agent.
-5. **Fallback** – Provide plain-text instructions when the UI cannot render.
-
-### Templates & examples
-
-- [Single-select deployment decision](examples.md#single-select-template)
-- [Multi-select diagnostics checklist](examples.md#multi-select-variation)
-- [Follow-up drill-down prompt](examples.md#follow-up-prompt-for-deeper-detail)
-
-## Best Practices
-- Reduce context switching by putting questions, options, and follow-up actions on one screen.
-- Options are sorted according to comparison criteria such as risk and priority.
-- If safety measures are required, such as approval/suspension/cancellation, attach a warning message.
-- Submission results are recorded for reuse in the Sync step or reports.
-
-## Inputs
-- Decision scenarios and candidate options collected by Alfred workflows.
-- Definition of follow-up actions for each option (next command, subagent, TODO, etc.).
-
-## Outputs
-- AskUserQuestion block and choice → follow-up action mapping.
-- Summary of selection results and TODO items that require further confirmation.
-
-## Failure Modes
-- Ambiguous or duplicate options cause selection delays.
-- It is difficult to judge priorities when all options look the same.
-- AskUserQuestion does not work in a TUI-inactive environment, so plan B is needed.
-
-## Dependencies
-- Works with the main `/alfred:*` command and schedules follow-up actions with TodoWrite when necessary.
-- When combined with moai-foundation-ears · moai-foundation-tags, requirements → options → TAG records are connected.
-
-## References
-- Jakubovic, J. "Designing Effective CLI Dialogs." ACM Queue, 2021.
-- NCurses. "Programming Guide." https://invisible-island.net/ncurses/man/ncurses.3x.html (accessed 2025-03-29).
-
-## Changelog
-- 2025-03-29: Added input/output/failure response to Alfred-specific skills.
-
-## Works well with
-
-- `moai-foundation-ears` – Combine structured requirement patterns with menu-driven confirmations.
-- `moai-alfred-git-workflow` – Use menus to choose branch/worktree strategies.
-- `moai-alfred-code-reviewer` – Capture reviewer focus areas through guided selection.
-
-## Examples
-```markdown
-- In the `/alfred:1-plan` step, user priorities are collected and the results are written to the PLAN board.
-- Check user approval before high-risk operations while running `/alfred:2-run`.
-```

moai_adk/templates/.claude/skills/moai-alfred-tui-survey/examples.md

@@ -1,62 +0,0 @@
-# Alfred TUI Survey Examples
-
-## Single-select template
-
-```typescript
-AskUserQuestion({
-  questions: [{
-    header: "Decision point: Deployment Strategy",
-    question: "How should we roll out the new release?",
-    options: [
-      {
-        label: "Canary release",
-        description: "Gradually roll out to a small user segment; monitor metrics first."
-      },
-      {
-        label: "Blue/Green",
-        description: "Keep the current version live while preparing the new stack in parallel."
-      },
-      {
-        label: "Full deploy",
-        description: "Immediate production rollout after smoke tests succeed."
-      }
-    ],
-    multiSelect: false
-  }]
-})
-```
-
-## Multi-select variation
-
-```typescript
-AskUserQuestion({
-  questions: [{
-    header: "Select diagnostics to run",
-    question: "Which checks should run before proceeding?",
-    options: [
-      { label: "Unit tests", description: "Fast verification for core modules." },
-      { label: "Integration tests", description: "Service-level interactions and DB calls." },
-      { label: "Security scan", description: "Dependency vulnerability audit." }
-    ],
-    multiSelect: true
-  }]
-})
-```
-
-## Follow-up prompt for deeper detail
-
-```typescript
-if (selection.includes("Integration tests")) {
-  AskUserQuestion({
-    questions: [{
-      header: "Integration test scope",
-      question: "Which environment should host integration tests?",
-      options: [
-        { label: "Staging", description: "Use the shared staging cluster." },
-        { label: "Ephemeral env", description: "Provision a one-off test environment." }
-      ],
-      multiSelect: false
-    }]
-  })
-}
-```

moai_adk/templates/.claude/skills/moai-claude-code/SKILL.md

@@ -1,94 +0,0 @@
----
-
-name: moai-claude-code
-description: Scaffolds and audits Claude Code agents, commands, skills, plugins, and settings with production templates. Use when configuring or reviewing Claude Code automation inside MoAI workflows.
-allowed-tools:
-  - Read
-  - Write
-  - Edit
----
-
-# MoAI Claude Code Manager
-
-## Skill Metadata
-| Field | Value |
-| ----- | ----- |
-| Allowed tools | Read (read_file), Write (write_file), Edit (edit_file) |
-| Auto-load | SessionStart (Claude Code bootstrap) |
-| Trigger cues | Agent/command/skill/plugin/settings authoring, Claude Code environment setup. |
-
-Create and manage Claude Code's five core components according to official standards.
-
-## Components covered
-
-- **Agents** `.claude/agents/` — Persona, tools, and workflow definition
-- **Commands** `.claude/commands/` — Slash command entry points
-- **Skills** `.claude/skills/` — Reusable instruction capsules
-- **Plugins** `settings.json › mcpServers` — MCP integrations
-- **Settings** `.claude/settings.json` — Tool permissions, hooks, session defaults
-
-## Reference files
-
-- `reference.md` — Writing guide and checklist
-- `examples.md` — Sample completed artifacts
-- `templates/` — Markdown/JSON skeleton for five components
-- `scripts/` — settings validation and template integrity check scripts
-
-## Workflow
-
-1. Analyze user requests to determine required components (Agent/Command/Skill/Plugin/Settings).
-2. Copy the stub from `templates/` and replace the placeholders to suit your project context.
-3. If necessary, run the `scripts/` verifier to check required fields, permissions, and follow-up links.
-
-## Guardrails
-
-- Maintain minimum privileges and progressive disclosure in line with Anthropic official guidelines.
-- Rather than modifying the template directly, only update the hook/field guided by reference.md.
-- The created files and settings.json are included in Git version management and a change history is left.
-
-**Official documentation**: https://docs.claude.com/en/docs/claude-code/skills  
-**Version**: 1.0.0
-
-## Examples
-```markdown
-- Create a spec-builder agent and the /alfred command set in a new project.
-- Review existing settings.json to update allowable tools and hook configurations.
-```
-
-## Best Practices
-- The output template is designed to be idempotent so that it is safe even when reapplied.
-- Detailed procedures are separated into reference.md (writing guide) and examples.md (sample artifacts) and loaded only when necessary.
-
-## When to use
-- Activates when someone asks to scaffold or audit Claude Code components.
-- When bootstrapping a Claude Code configuration to a new project.
-- When reexamining existing agents/commands/skills/plug-ins to meet standards.
-- When verification of settings is required in initialization workflows such as `/alfred:0-project`.
-
-## What it does
-- Create and update the five core components as official templates.
-- Verify accepted tools, model selection, and progressive disclosure links.
-- Provides reusable stubs and verification procedures through templates/·scripts/ resources.
-
-## Inputs
-- User configuration requests (e.g. “add new command”, “review settings.json”) and current `.claude/` directory state.
-- Project-specific template requirements or security/permissions policies.
-
-## Outputs
-- Official Markdown definition file under `.claude/agents|commands|skills/`.
-- A summary of `.claude/settings.json` that reflects the latest settings and allowed tools and follow-up TODOs.
-
-## Failure Modes
-- If the template path or placeholder is different from the latest version, the result may be damaged.
-- The settings.json permission policy may conflict with project rules or may block verification script execution.
-
-## Dependencies
-- When used with cc-manager, doc-syncer, and moai-foundation-git, the creation → verification → distribution flow is completed.
-- You must have versioned resources in the templates/ and scripts/ directories for automation to work properly.
-
-## References
-- Anthropic. "Claude Code Style Guide." https://docs.claude.com/ (accessed 2025-03-29).
-- Prettier. "Opinionated Code Formatter." https://prettier.io/docs/en/ (accessed 2025-03-29).
-
-## Changelog
-- 2025-03-29: Added best practice structure to Claude code formatting skill.