@haposoft/cafekit 0.7.22 → 0.7.24

package/README.md

@@ -1,94 +1,28 @@
 # @haposoft/cafekit
 
-> Spec-Driven Development workflow for AI coding assistants
+> Claude Code-first spec-driven workflow and runtime bundle for AI coding assistants.
 
-[![Version](https://img.shields.io/badge/version-0.3.1-blue.svg)](https://github.com/haposoft/cafekit)
+[![Version](https://img.shields.io/badge/version-0.7.23-blue.svg)](https://github.com/haposoft/cafekit)
 [![License](https://img.shields.io/badge/license-MIT-green.svg)](LICENSE)
-[![Claude Code](https://img.shields.io/badge/Claude%20Code-Supported-orange.svg)](https://claude.ai/code)
-[![Antigravity](https://img.shields.io/badge/Antigravity-Supported-purple.svg)](https://github.com/google/antigravity)
+[![Claude%20Code](https://img.shields.io/badge/Claude%20Code-Primary-orange.svg)](https://claude.ai/code)
 
 ## Overview
 
-CafeKit is a **multi-platform** CLI tool that installs a structured workflow for AI coding assistants. It helps teams move from idea to implementation systematically using natural language commands.
-
-**Supported Platforms:**
-| Platform | Status | Installation Path |
-|----------|--------|-------------------|
-| [Claude Code](https://claude.ai/code) | ✅ Supported | `.claude/commands/` |
-| [Antigravity](https://antigravity.google/) | ✅ Supported | `.agent/workflows/` |
-| Cursor | 🔮 Planned | 🔮 Planned |
-| GitHub Copilot | 🔮 Planned | TBD |
-| Windsurf | 🔮 Planned | TBD |
-
-**What it does:**
-- Installs workflow commands into your AI editor command/workflow folder
-- Enables spec-driven development with clear phase separation
-- Creates living documentation for every feature
-- Works with zero configuration
-- **Idempotent by default** - Safe to re-run, skips existing files
-- **Upgrade mode available** - Use `--upgrade` (or `--force`) to refresh managed templates
-
-**What it doesn't do:**
-- Generate code (commands guide AI to help you write code)
-- Require configuration (zero-config installation)
-- Lock you to a specific AI editor
-
-## Features
-
-- **🎯 Multi-platform** - Works with Claude Code, Antigravity, and future AI editors
-- **📋 Spec-first workflow** - `spec-init - spec-requirements - spec-design - spec-tasks` + `code - test - review`
-- **📝 Documentation workflow** - `/docs init` and `/docs update` for project documentation
-- **⚡ Zero-config** - Works out of the box with sensible defaults
-- **🔄 Idempotent** - Safe to re-run, skips existing files
-- **📦 No global install** - Use directly with `npx`
-- **🚀 Future-proof** - Easy to add support for new AI editors
-
-## Claude Code Statusline (Claude Code Only)
-
-CafeKit automatically installs an enhanced statusline for Claude Code that provides real-time session context.
-
-**What it shows:**
-- **Context usage** - Percentage and token count (e.g., `23% 45K/200K`)
-- **Session timer** - Elapsed time since session start
-- **Git status** - Current branch and dirty state indicator
-- **Active agents** - Count of running subagents
-- **Todo items** - Count of pending tasks
-
-**Installation:**
-- Automatically installed when running `npx @haposoft/cafekit` in a Claude Code project
-- Merges with existing `settings.json` configuration without overwriting user settings
-- Safe to re-run - preserves non-CafeKit statusline configurations
-- Upgrade mode (`--upgrade`) refreshes managed runtime files
-
-**Configuration:**
-The statusline is configured via `.claude/settings.json`:
-```json
-{
-  "statusLine": {
-    "type": "command",
-    "command": "node \"$CLAUDE_PROJECT_DIR/.claude/status.cjs\"",
-    "padding": 0
-  }
-}
-```
-
-**Runtime files installed:**
-- `.claude/status.cjs` - Main statusline script
-- `.claude/hooks/session.cjs` - Session initialization hook
-- `.claude/hooks/agent.cjs` - Subagent context injection hook
-- `.claude/hooks/usage.cjs` - Usage tracking hook
-- `.claude/hooks/lib/*.cjs` - Shared utilities (color, parser, git, config, etc.)
+CafeKit installs a structured workflow into Claude Code so the assistant can move cleanly from:
 
-**Note:** This feature is Claude Code exclusive and not available for Antigravity.
+```text
+Idea -> Spec -> Design -> Task Files -> Implementation -> Test -> Review
+```
 
-## Installation
+This package currently focuses on the Claude Code runtime:
+- installs CafeKit skills under `.claude/skills/`
+- installs supporting agents under `.claude/agents/`
+- installs runtime hooks, statusline, and managed settings
+- merges Claude settings safely on re-run
 
-### Prerequisites
+Antigravity remains in the repository as a legacy path, but Claude Code is the primary supported runtime for this release.
 
-- Node.js 18+
-- An AI coding assistant (Claude Code or Antigravity)
-
-### Install Workflow
+## Install
 
 Run in your project root:
 
@@ -96,833 +30,118 @@ Run in your project root:
 npx @haposoft/cafekit
 ```
 
-The installer will:
-1. **Auto-detect** your AI editor configuration (`.claude/` or `.agent/`)
-2. **Prompt** you to select platform if not detected
-3. **Copy** workflow commands to the appropriate directory
-4. **Install** shared skills for spec-driven development
-5. **[Claude Code only]** Install unprefixed skill directories (`spec-init`, `spec-requirements`, `spec-design`, `spec-tasks`, `code`, `test`, `review`) that expose `hapo:`-prefixed skill names
-6. **Ensure dependencies** for `code - test - review` by installing missing command/agent templates
-
-Installer modes:
-- **Default install mode**: `npx @haposoft/cafekit` (skip existing files)
-- **Upgrade mode**: `npx @haposoft/cafekit --upgrade` (overwrite managed templates)
-- **Alias**: `--force` / `-f` behaves the same as `--upgrade`
-
-**Example output (Claude Code):**
-```
-CafeKit Installer v0.3.12
-========================================
-
-Installing for: Claude Code
-Mode: install (skip existing files)
-
-Claude Code (.claude/)
-----------------------------------------
-✓ Skill installed: specs
-✓ Skill installed: spec-init
-✓ Skill installed: spec-requirements
-✓ Skill installed: spec-design
-✓ Skill installed: spec-tasks
-✓ Skill installed: code
-✓ Skill installed: test
-✓ Skill installed: review
-✓ Copied: spec-init.md
-✓ Copied: spec-requirements.md
-...
-
-╔════════════════════════════════════════════════════════╗
-║         Installation Complete!                         ║
-╚════════════════════════════════════════════════════════╝
-
-  Installed Skills:   Yes ✓
-
-Next steps:
-  1. Start your AI editor
-
-  For Claude Code:
-     Run: /spec-init <feature-name>
-     Or use skill: /hapo:spec-init <feature-description>
-
-  2. Follow the workflow: requirements - design - tasks - code - test - review
-
-Documentation: https://github.com/haposoft/cafekit
-```
-
-## Workflows
-
-CafeKit provides two workflow categories:
-
-### 1. Spec-Driven Development Workflows (`spec-*`)
-
-**Purpose:** Structured feature development from idea to implementation
-
-**Claude Code Commands:**
-| Command | Purpose | Phase |
-|---------|---------|-------|
-| `/spec-init` | Initialize new feature spec | 1 |
-| `/spec-requirements` | Generate EARS requirements | 2 |
-| `/spec-design` | Create technical design | 3 |
-| `/spec-validate` | Validate design decisions via interview | 4 |
-| `/spec-tasks` | Break down into tasks | 5 |
-| `/code` | Implement tasks from spec artifacts | 6 |
-| `/test` | Run tests for implemented changes | 7 |
-| `/review` | Review code quality and risks | 8 |
-| `/spec-status` | Check progress | 9 |
-
-**Antigravity Workflows:**
-| Workflow | Purpose | Phase |
-|----------|---------|-------|
-| `/spec-init` | Initialize new feature spec | 1 |
-| `/spec-requirements` | Generate EARS requirements | 2 |
-| `/spec-design` | Create technical design | 3 |
-| `/spec-validate` | Validate design decisions via interview | 4 |
-| `/spec-tasks` | Break down into tasks | 5 |
-| `/code` | Implement tasks from spec artifacts | 6 |
-| `/test` | Run tests for implemented changes | 7 |
-| `/review` | Review code quality and risks | 8 |
-| `/spec-status` | Check progress | 9 |
-
-**Generated Files:**
-```
-.specs/
-└── feature-name/
-    ├── spec.json
-    ├── requirements.md
-    ├── design.md
-    └── tasks/
-```
-
----
-
-### 2. Documentation Workflows (`docs`)
-
-**Purpose:** Project documentation management and maintenance
-
-**Claude Code Commands:**
-| Command | Purpose |
-|---------|---------|
-| `/docs init` | Initialize comprehensive documentation |
-| `/docs update` | Update docs after code changes |
-
-**Antigravity Workflows:**
-| Workflow | Purpose |
-|----------|---------|
-| `/docs-init` | Initialize comprehensive documentation |
-| `/docs-update` | Update docs after code changes |
-
-**Generated Files:**
-```
-docs/
-├── codebase-summary.md
-├── project-overview-pdr.md
-├── code-standards.md
-├── system-architecture.md
-├── design-guidelines.md
-├── deployment-guide.md
-└── project-roadmap.md
-```
-
----
-
-## Quick Start
-
-### A. Spec Workflow - Building User Authentication
-
-```bash
-# Step 1: Initialize spec
-/spec-init User authentication with JWT and refresh tokens
-
-# AI creates:
-# - .specs/user-authentication/
-# - .specs/user-authentication/spec.json
-# - .specs/user-authentication/requirements.md
-
-# Step 2: Generate requirements
-/spec-requirements user-authentication
-
-# AI analyzes codebase and generates:
-# - Functional requirements (EARS format)
-# - Acceptance criteria
-# - Constraints
-
-# Step 3: Create design
-/spec-design user-authentication
-
-# AI generates:
-# - Architecture decisions
-# - API endpoints
-# - Database schema
-# - Component structure
-
-# Step 4: Validate design decisions (recommended for medium/high-risk features)
-/spec-validate user-authentication
-
-# AI confirms:
-# - Critical trade-offs and assumptions
-# - Risk-sensitive defaults
-# - Follow-up actions before task generation
-
-# Step 5: Break down tasks
-/spec-tasks user-authentication
-
-# AI creates:
-# - Prioritized task list
-# - Dependencies
-# - Estimated complexity
-
-# Step 6: Code
-/code user-authentication
-
-# Step 7: Test
-/test
-
-# Step 8: Review
-/review
-
-# Step 9: Check status
-/spec-status user-authentication
-
-# AI reports:
-# - Progress (5/8 tasks complete)
-# - Blockers
-# - Next steps
-```
-
-## Workflow Overview
-
-### Workflow Diagram
-
-```
-Idea - /spec-init - /spec-requirements - /spec-design - /spec-validate - /spec-tasks - /code - /test - /review - /spec-status
-       |<----------------------------------------------|-----------------------------|
-                                                      |
-                                                   iterate
-```
-
-### 1. /spec-init
-
-**Purpose:** Initialize a new specification from a project description.
-
-**When to use:** Starting a new feature, enhancement, or component.
-
-**What it does:**
-- Generates feature name from description
-- Creates `.specs/[feature-name]/` directory
-- Initializes `spec.json` metadata
-- Creates `requirements.md` template
-
-**Example:**
-```bash
-/spec-init Dark mode toggle with user preference persistence
-
-# Creates:
-# .specs/dark-mode-toggle/
-#   ├── spec.json
-#   └── requirements.md (with initial description)
-```
-
-**Tips:**
-- Be specific (5+ words describing what to build)
-- Avoid vague terms like "improve" or "better" without context
-- Include key nouns (e.g., "login form", "dashboard chart")
-
----
-
-### 2. /spec-requirements
-
-**Purpose:** Generate comprehensive requirements in EARS format.
-
-**When to use:** After initialization, before design phase.
-
-**What it does:**
-- Analyzes existing codebase for related code
-- Loads project steering context (`.specs/steering/`)
-- Generates functional requirements with acceptance criteria
-- Documents constraints and assumptions
+Refresh managed files:
 
-**Example:**
 ```bash
-/spec-requirements dark-mode-toggle
-
-# Updates requirements.md with:
-# - Functional Requirements (EARS format)
-#   - "When user clicks toggle, system shall switch theme"
-#   - "When theme changes, system shall persist preference"
-# - Acceptance Criteria
-# - Constraints (browser support, accessibility)
-```
-
-**EARS format example:**
-```markdown
-## Functional Requirements
-
-### FR-1: Theme Toggle
-When user clicks the theme toggle button,
-the system shall switch between light and dark themes
-and update UI within 200ms.
-
-**Acceptance Criteria:**
-- Toggle state reflects current theme
-- Animation is smooth (no flash)
-- All components respect theme
-```
-
----
-
-### 3. /spec-design
-
-**Purpose:** Create detailed technical design from requirements.
-
-**When to use:** After requirements are finalized, before task breakdown.
-
-**What it does:**
-- Generates architecture decisions
-- Defines API contracts (if needed)
-- Creates database schema (if needed)
-- Specifies component structure
-- Documents technology choices
-
-**Example:**
-```bash
-/spec-design dark-mode-toggle
-
-# Creates design.md with:
-# - Architecture: Context API for theme state
-# - Components: ThemeToggle.tsx, ThemeProvider.tsx
-# - Storage: localStorage with 'theme' key
-# - CSS: Tailwind dark: variants
-```
-
-**Output structure:**
-```markdown
-# Design Specification
-
-## Architecture Decisions
-- Use React Context for global theme state
-- Tailwind CSS dark mode (class strategy)
-
-## Component Structure
-- ThemeProvider (context wrapper)
-- ThemeToggle (button component)
-- useTheme hook (consume context)
-
-## Data Flow
-User clicks -> dispatch action -> update context -> localStorage -> re-render
-```
-
----
-
-### 4. /spec-validate
-
-**Purpose:** Validate design assumptions, trade-offs, and risk-sensitive decisions before task generation.
-
-**When to use:** After `/spec-design`, especially for medium/high-risk features or external integrations.
-
-**What it does:**
-- Interviews critical decision points using structured questions
-- Appends validation log to `research.md`
-- Updates validation metadata in `spec.json`
-- Recommends follow-up actions before task breakdown
-
-**Example:**
-```bash
-/spec-validate dark-mode-toggle
-```
-
----
-
-### 5. /spec-tasks
-
-**Purpose:** Break design into implementable tasks.
-
-**When to use:** After design is approved, before implementation.
-
-**What it does:**
-- Generates prioritized task list
-- Identifies dependencies between tasks
-- Estimates complexity (simple/medium/complex)
-- Creates execution order
-
-**Example:**
-```bash
-/spec-tasks dark-mode-toggle
-
-# Creates tasks.md with:
-# Task #1: Create ThemeContext and Provider (simple)
-# Task #2: Build ThemeToggle button component (simple)
-# Task #3: Implement localStorage persistence (medium)
-# Task #4: Add dark mode CSS variables (medium)
-# Task #5: Test theme switching across pages (medium)
-```
-
-**Task format:**
-```markdown
-## Task #1: Create ThemeContext
-- **Priority:** High
-- **Complexity:** Simple
-- **Dependencies:** None
-- **Files:** src/context/ThemeContext.tsx
-- **Description:** Set up React Context with theme state
-- **Status:** pending
-```
-
----
-
-### 5. /code
-
-**Purpose:** Implement a specific task from the task list.
-
-**When to use:** After `/spec-tasks`, to implement approved tasks from spec artifacts.
-
-**What it does:**
-- Reads spec artifacts (`tasks.md`, `design.md`, `requirements.md`)
-- Implements approved work with minimal scope
-- Hands off to `/test` and `/review`
-- Reports blockers and next actions
-
-**Example:**
-```bash
-# Implement from approved spec tasks
-/code dark-mode-toggle
-
-# Then run quality gates
-/test
-/review
-```
-
-**Iteration pattern:**
-```bash
-# Repeat the quality loop after each coding pass
-/code dark-mode-toggle
-/test
-/review
-```
-
----
-
-### 6. /test
-
-**Purpose:** Run tests after `/code`.
-
-**When to use:** Immediately after each coding pass.
-
-**Example:**
-```bash
-/test
-```
-
----
-
-### 7. /review
-
-**Purpose:** Review recent changes for quality and risk.
-
-**When to use:** After `/test` passes.
-
-**Example:**
-```bash
-/review
-```
-
----
-
-### 8. /spec-status
-
-**Purpose:** Check progress and next steps.
-
-**When to use:** Anytime during or after implementation.
-
-**What it does:**
-- Reads `tasks.md` for completion status
-- Calculates progress percentage
-- Identifies blockers or pending tasks
-- Suggests next actions
-
-**Example:**
-```bash
-/spec-status dark-mode-toggle
-
-# Output:
-# Progress: 3/5 tasks completed (60%)
-# Completed: #1, #2, #3
-# Pending: #4, #5
-# Next: Implement Task #4 (Add dark mode CSS variables)
-# Blockers: None
-```
-
-**Status indicators:**
-- `pending` - Not started
-- `in-progress` - Partially implemented
-- `completed` - Finished
-- `blocked` - Waiting on dependency
-
----
-
-### B. Docs Workflow - Initialize Project Documentation
-
-**Claude Code:**
-```bash
-# Initialize documentation
-/docs init
-
-# AI creates:
-# - docs/codebase-summary.md
-# - docs/project-overview-pdr.md
-# - docs/code-standards.md
-# - docs/system-architecture.md
-# - docs/design-guidelines.md
-# - docs/deployment-guide.md
-# - docs/project-roadmap.md
-# - .agent/rules/AGENTS.md (for Antigravity) or CLAUDE.md (for Claude Code)
-
-# Update after code changes
-/docs update
-```
-
-**Antigravity:**
-```bash
-# Initialize documentation
-/docs-init
-
-# AI creates the same 7 docs files + AGENTS.md
-
-# Update after code changes
-/docs-update
+npx @haposoft/cafekit --upgrade
 ```
 
----
+Requirements:
+- Node.js 18+
+- Claude Code project with a `.claude/` directory, or choose Claude when prompted
 
-## File Structure
+## What Gets Installed
 
-### Platform-Specific Installation
+Claude Code install targets:
 
-**Claude Code** (`.claude/`):
-```
+```text
 .claude/
-├── commands/
-│   ├── spec-init.md          # Spec workflows
-│   ├── spec-requirements.md
-│   ├── spec-design.md
-│   ├── spec-tasks.md
-│   ├── code.md
-│   ├── test.md
-│   ├── review.md
-│   ├── spec-status.md
-│   └── docs.md               # Docs workflows
-└── skills/
-    ├── specs/
-    ├── impact-analysis/
-    ├── spec-init/
-    ├── spec-requirements/
-    ├── spec-design/
-    ├── spec-tasks/
-    ├── code/
-    ├── test/
-    └── review/
-```
-
-**Antigravity** (`.agent/`):
-```
-.agent/
-├── workflows/                   # Antigravity workflows (hyphen naming)
-│   ├── spec-init.md
-│   ├── spec-requirements.md
-│   ├── spec-design.md
-│   ├── spec-tasks.md
-│   ├── code.md
-│   ├── test.md
-│   ├── review.md
-│   ├── spec-status.md
-│   ├── docs-init.md            # Docs workflows
-│   └── docs-update.md
 ├── skills/
-│   ├── specs/
-│   └── impact-analysis/
-└── rules/
-    └── GEMINI.md               # System rules (always_on)
+├── agents/
+├── hooks/
+├── status.cjs
+├── runtime.json
+├── settings.json
+└── CLAUDE.md
 ```
 
-**Command Naming:**
-- **Claude Code:** Uses hyphens (`/spec-init`, `/docs init`)
-- **Antigravity:** Uses hyphens (`/spec-init`, `/docs-init`)
-
-### Generated Specs Directory
+Managed runtime features include:
+- Claude Code statusline
+- session and subagent hooks
+- rule/context injection
+- spec state awareness
+- safe settings merge on reinstall
 
-Specs are created in `.specs/` (shared across all platforms):
-```
-.specs/
-├── steering/              # Project-wide conventions
-│   ├── tech.md
-│   ├── structure.md
-│   └── product.md
-└── feature-name/          # Individual feature specs
-    ├── spec.json
-    ├── requirements.md
-    ├── design.md
-    ├── research.md
-    └── tasks/
-        └── sprint-1.md
-```
+## Core Skills
 
----
+CafeKit ships many skills, but the main release surface is:
 
-## FAQ
+- `/hapo:specs <feature-description>`: create or resume a structured spec workflow
+- `/hapo:develop <feature-name>`: implement from approved spec artifacts
+- `/hapo:test [scope|--full]`: run verification and return a structured verdict
+- `/hapo:code-review [scope|--pending]`: adversarial review focused on correctness, regressions, and security
+- `/hapo:generate-graph <diagram request>`: generate technical SVG/PNG diagrams
 
-### Q: Which AI editors are supported?
+Common companion skills bundled in this package include `inspect`, `impact-analysis`, `research`, `ai-multimodal`, `frontend-development`, `backend-development`, and `react-best-practices`.
 
-A: Currently supports:
-- ✅ **Claude Code** - `.claude/commands/`
-- ✅ **Antigravity** - `.agent/workflows/`
-
-Planned for future:
-- 🔮 **Cursor** - `.cursor/commands/`
-- 🔮 **GitHub Copilot** - TBD
-- 🔮 **Windsurf** - TBD
+## Quick Start
 
-### Q: Can I use multiple AI editors on the same project?
+Create a new spec:
 
-A: Yes! You can install CafeKit for multiple platforms:
 ```bash
-npx @haposoft/cafekit
-# Select "Both" when prompted
+/hapo:specs Build a Google Meet transcript extension with AI summaries
 ```
 
-This installs commands to both `.claude/` and `.agent/`. The `.specs/` directory is shared, so your specifications work across all editors.
+Implement the whole feature:
 
-### Q: Can I customize the workflow steps?
-
-A: Yes. Edit the `.md` files in `.claude/commands/` (Claude Code) or `.agent/workflows/` (Antigravity) after installation.
-
-### Q: Is it safe to re-run `npx @haposoft/cafekit`?
-
-A: Yes. Default mode is idempotent and skips existing files.
-
-### Q: How do I update existing installed templates to the latest version?
-
-A: Run installer in upgrade mode:
 ```bash
-npx @haposoft/cafekit --upgrade
-# or
-npx @haposoft/cafekit --force
+/hapo:develop meet-transcript-mvp
 ```
-This overwrites files managed by the installer (commands/workflows, managed dependencies, and routing/rules templates).
 
-### Q: What language is the spec generated in?
+Implement one specific task file:
 
-A: Defaults to English.
-
-### Q: Can I use this for multiple features simultaneously?
-
-A: Yes. Each spec has its own directory in `.specs/`. Work on multiple in parallel.
-
-### Q: Why are the command names different between Claude Code and Antigravity?
-
-A: Each platform has different conventions:
-- **Claude Code** uses hyphens: `/spec-init`, `/docs update`
-- **Antigravity** uses hyphens: `/spec-init`, `/docs-init`
-
-The functionality is identical, only the naming convention differs.
-
----
-
-## Troubleshooting
-
-### "No platform configuration detected"
-
-**Cause:** Neither `.claude/` nor `.agent/` folder exists in your project.
-
-**Solution:**
 ```bash
-# Option 1: Create Claude Code folder
-mkdir -p .claude/commands
-
-# Option 2: Create Antigravity folder
-mkdir -p .agent/workflows
-
-# Option 3: Run installer and it will prompt you to select platform
-npx @haposoft/cafekit
+/hapo:develop meet-transcript-mvp task-R0-02-extension-scaffold-dashboard-skeleton.md
 ```
 
-### "Command not found" after installation
+Run tests and review:
 
-**Claude Code:**
-- Commands load automatically
-- If not visible, restart Claude Code extension
-- Verify: `ls .claude/commands/spec-*.md`
-
-**Antigravity:**
-- Workflows load automatically when IDE opens the project
-- Verify: `ls .agent/workflows/*.md`
-- Check `GEMINI.md` is in `.agent/rules/` for system rules
-
----
-
-## Best Practices
-
-### 1. Phase Separation
-
-Complete each phase before moving to the next:
-- Don't skip requirements (even for "simple" features)
-- Finalize design before creating tasks
-- Implement tasks sequentially (respect dependencies)
-
-### 2. Incremental Implementation
-
-Break large features into smaller specs:
 ```bash
-# Instead of:
-/spec-init Complete e-commerce system
-
-# Do:
-/spec-init Product listing page
-/spec-init Shopping cart functionality
-/spec-init Checkout flow
-```
-
-### 3. Steering Context
-
-For projects with shared conventions:
-1. Create `.specs/steering/` directory
-2. Add files like `tech.md`, `structure.md`, `product.md`
-3. All specs will automatically reference these
-
-Example `.specs/steering/tech.md`:
-```markdown
-# Technology Standards
-
-- Frontend: React 18 + TypeScript
-- Styling: Tailwind CSS v4
-- State: Zustand
-- Testing: Vitest + Playwright
+/hapo:test --full
+/hapo:code-review --pending
 ```
 
-### 4. Version Control
+Generate a diagram:
 
-Commit after each phase:
 ```bash
-/spec-init feature-name
-git add .specs/feature-name/
-git commit -m "spec: init feature-name"
-
-/spec-requirements feature-name
-git add .specs/feature-name/requirements.md
-git commit -m "spec: requirements for feature-name"
-
-# ... continue for design, tasks, impl
+/hapo:generate-graph Draw a sequence diagram for auth flow between browser, API, and database
 ```
 
-### 5. Task Granularity
+## Spec Artifacts
 
-Good task size: 30-60 minutes of work.
-
-Too large:
-```markdown
-Task #1: Build entire authentication system
-```
+CafeKit's current spec workflow writes artifacts under:
 
-Better:
-```markdown
-Task #1: Create User model and migration
-Task #2: Implement JWT generation
-Task #3: Add login endpoint
-Task #4: Add protected route middleware
+```text
+specs/<feature-name>/
+├── spec.json
+├── requirements.md
+├── research.md
+├── design.md
+└── tasks/
+    ├── task-R0-01-*.md
+    ├── task-R1-01-*.md
+    └── ...
 ```
 
----
-
-## Contributing
-
-- **Report issues:** https://github.com/haposoft/cafekit/issues
-- **Suggest improvements:** Open a GitHub Discussion
-
----
-
-## Support
-
-- **GitHub:** https://github.com/haposoft/cafekit
-- **Issues:** https://github.com/haposoft/cafekit/issues
-- **Discussions:** https://github.com/haposoft/cafekit/discussions
-
----
-
-## Changelog
-
-### [0.3.1] - 2026-03-13
-
-#### Changed
-- **Renamed package** from `@haposoft/cafekit-spec` to `@haposoft/cafekit`
-- Deprecated the old `@haposoft/cafekit-spec` package
-- Updated repository and homepage URLs to point to the official `haposoft` organization
+The active workflow expects:
+- `spec.json` to hold state, approvals, validation, and `task_files`
+- design to define canonical contracts
+- each task file to carry completion criteria and verification evidence
 
-### [0.2.2] - 2026-02-25
+## Release Notes For 0.7.23
 
-#### Changed
-- Added `scope_lock` contract to spec initialization metadata (`source`, `in_scope`, `out_of_scope`, `expansion_policy`)
-- Updated `/spec-requirements`, `/spec-design`, `/spec-validate`, and `/spec-tasks` to enforce scope lock across the full spec lifecycle
-- Reduced default expansion in `/spec-design`: uncertain cases now default to light discovery and only escalate to full on explicit triggers
-- Clarified steering usage in requirements phase: steering is constraints-only and must not introduce new capability domains
-- Added task-generation guardrail: every task must map to valid in-scope numeric requirement IDs
-- Added installer sync for specs template files in Claude mode to reduce stale runtime copy drift
+This release is centered on Claude Code:
+- tightened `hapo:specs` state integrity and task finalization rules
+- tightened `hapo:develop` definition-of-done and evidence-based quality gates
+- bundled `hapo:generate-graph`
+- cleaned Claude installer expectations so it no longer looks for removed Claude artifacts
 
-#### Regression Guard (installer CLI upgrade mode)
-Use this checklist after running `npx @haposoft/cafekit --upgrade`:
-- Run `/spec-init` with an installer-CLI-scoped feature description and verify `spec.json.scope_lock` is populated
-- Run `/spec-requirements`, `/spec-design`, `/spec-validate`, `/spec-tasks` in order
-- Verify requirements/design/tasks do **not** introduce API/mobile/DynamoDB domains unless explicitly approved
-- Verify task entries reference valid in-scope numeric requirement IDs only
-- If scope expansion is explicitly approved during validation, verify `scope_lock` updates are reflected in `spec.json`
+## Documentation
 
-### [0.1.7] - 2026-02-24
-
-#### Changed
-- Unified workflow naming around hyphens for Antigravity workflows
-- Replaced `spec-impl` in primary flow with `code - test - review`
-- Installer now ensures dependency templates for `code/test/review` commands and required agents
-- Updated docs workflow naming to `/docs-init` and `/docs-update` for Antigravity
-
-### [0.1.5] - 2026-02-11
-
-#### Added
-- **Documentation workflow for Antigravity** - New `/docs-init` and `/docs-update` workflows
-- **GEMINI.md rule file** - Auto-installs `.agent/rules/GEMINI.md` with system rules for Antigravity
-- **Full Antigravity support** - Documentation commands now work on both Claude Code and Antigravity
-- **AGENTS.md auto-generation** - Created automatically when running `/docs-init` or `/docs-update`
-
-### [0.1.2] - 2026-02-04
-
-#### Changed
-- **Multi-platform support** - Renamed from "Claude Code only" to "AI coding assistants"
-- Updated documentation to reflect dual-platform support (Claude Code + Antigravity)
-- Improved platform detection and installation UX
-- Added platform compatibility matrix
-
-### [0.1.1] - 2026-02-03
-
-#### Added
-- Dual-platform installer supporting both Claude Code and Antigravity
-- Auto-detection of existing `.claude/` and `.agent/` configurations
-- Interactive platform selection when no configuration detected
-
-### [0.1.0] - 2026-02-02
-
-#### Added
-- Initial release of CafeKit workflow
-- Initial spec workflow foundation
-- Zero-config installation via npx
-- Idempotent file copying (safe to re-run)
-- Comprehensive documentation
-
----
+- Installation: https://cafekit.haposoft.com/docs/getting-started/installation
+- Quickstart: https://cafekit.haposoft.com/docs/getting-started/quickstart
+- Spec workflow: https://cafekit.haposoft.com/docs/guides/spec-workflow
 
 ## License
 
-MIT License - See [LICENSE](LICENSE) for details
-
----
-
-**Made with care by the Haposoft Team**
-
-<p align="center">
-  <sub>Multi-platform Workflow for AI Coding Assistants</sub>
-</p>
+MIT © Haposoft