Claude CLI Advanced Starter Pack (CCASP)



---



---

v2.8.0 Feature Update — Neovim UI Testing Framework + Appshell Layout

> Feb 2026 — New deterministic headless Neovim UI testing framework for the nvim-ccasp plugin with plenary.nvim-based test harness, UI driver/snapshot modules, appshell layout (VS Code-inspired zone-based UI), and a complete Claude workflow system with 6 commands, 3 agents, 2 hooks, and per-folder documentation.

$3

#### Neovim UI Testing Framework Deterministic headless testing for nvim-ccasp: - Test Harness — plenary.nvim with smoke, keymap, and layout tests - UI Driver — Automated UI traversal (open, navigate, close) — 534 lines - Layout Snapshots — Capture and diff window layouts — 512 lines - Runner Scripts — nvim-test.ps1, nvim-smoke.ps1, nvim-snapshot.ps1 - Headless Execution — CI-ready with minimal_init.lua

#### Appshell Layout (nvim-ccasp v1.5.0) VS Code / Discord-inspired zone-based UI: - Icon Rail — 3-column vertical navigation strip - Flyout — Expandable 35-column sidebar panel - Header/Footer — Tabline + status bar zones - Content Zone — Adaptive main editing area - Right Panel — Optional context/logs panel - Zone-Aware Panels — All panels respect appshell bounds

#### Claude Workflow System 6 new slash commands for UI testing: - /ui-test — Run full test suite - /ui-smoke — Quick smoke test - /ui-snapshot — Layout snapshot + diff - /ui-bug — Structured bug reporting - /ui-fix — Guided debug/fix workflow - /repo-guard — Pre-commit quality gate

#### Agents, Hooks & Documentation 3 agents + 2 hooks + per-folder docs: - ui-driver agent — Drives UI, validates windows - ui-debugger agent — Diagnoses and fixes bugs - nvim-test-runner agent — Runs tests, reports results - nvim-after-edit-test hook — Auto-smoke on .lua edit - agent-limit-compact hook — Limits to 3 concurrent agents - 15 doc files across 4 nvim-ui-testing/ subfolders

$3

`bash

Update to v2.8.0

npm update -g claude-cli-advanced-starter-pack

Sync new commands to your project

ccasp init
`

---



v2.8.1 — Getting Started Onboarding + Help System (nvim-ccasp) (click to expand)

> Feb 2026 — Interactive 8-page Getting Started guide with first-launch detection, ASCII diagrams, try-it actions, and a persistent Help system with 11 searchable wiki topics embedded offline.

- Getting Started Guide — 8-page onboarding with ASCII diagrams, keybinding tables, try-it actions
- First-Launch Detection — Auto-opens on first launch, remembers completion state
- Help System — 11 wiki topics with Telescope fuzzy search + vim.ui.select fallback
- Navigation — Page dots, n/p keys, 1-8 jump, h/l arrow support
- New Commands — :CcaspWelcome, :CcaspHelp, c?, cw
- Sidebar Integration — Help tab now shows Getting Started and Help links







v2.7.0 — Vision Multi-Instance & Decision Engine (click to expand)

> Feb 2026 — Vision Mode now supports multiple concurrent visions with a centralized registry, cross-process file locking, slug uniqueness enforcement, and an intelligent decision engine.

- Vision Registry: Centralized VISION_REGISTRY.json with atomic writes, auto-rebuild
- Decision Engine: Weighted scoring selects plan type (task-list / phase-dev-plan / roadmap / epic / vision-full)
- Slug Uniqueness: generateUniqueSlug() appends -2/-3 on collision
- New Commands: ccasp vision list, ccasp vision cleanup, /vision-list







v2.6.1 — Smart Todo List with AI Intent Routing (click to expand)

> Feb 2026 — New Smart Todo system: quick-add with auto-paraphrasing, interactive menu, AI-powered "Work on item" that suggests the best CCASP workflow (Vision/Epic/Roadmap/Phase-Dev/Task List), plus full Neovim integration with sidebar, floating panel, and Telescope picker.

- /todo-add — Quick-add with auto-paraphrasing to 5-6 word titles
- /todo — Interactive menu: list, navigate, work-on, complete, delete
- AI Intent Routing — Analyzes todo and suggests Vision/Epic/Roadmap/Phase-Dev/Task List/Just do it
- Neovim Integration — Sidebar section, floating panel, Telescope picker



---

🚀 v2.6.0 Feature Update — Smart Site Intel Dev-Scan System

> Feb 2026 — New developer-focused application scanner: per-route health monitoring, data-testid coverage tracking, git-diff incremental scanning, regression detection, and interactive dashboard with 12 MCP tools.

$3

#### 🔬 Smart Site Intel Dev-Scan Developer-focused per-route application scanning: - Route Catalog — AST-based route discovery with reverse file→route mapping - Config Reader — Reads tech-stack.json, .env, playwright.config automatically - Playwright Scanner — Auth, live DOM testid check, axe a11y, optional Lighthouse - Git Diff Integration — Incremental scanning with 30% threshold for full scan - State Persistence — Per-route scores, history (20 entries FIFO), diff tracking - Quick Check — Static testid coverage in <10s (no Playwright needed)

#### 📊 Dev Scan Dashboard Interactive web dashboard at http://localhost:3847: - Health Metrics — Overall score, total routes, testid coverage, last scan - Sortable Route Table — Sort by path, testid, a11y, perf, health grade - Color-Coded Grades — A (green) through F (red) per-route health - Diff Display — Improvements (green) and regressions (red) - Trend Sparklines — Inline SVG health history per route - Filter Controls — All routes, failing (<70%), regressions only - 7 New API Endpoints — state, summary, routes, diffs, history, testid-coverage

#### 🔌 3 New MCP Tools (12 Total) Expanded site-intel MCP server: - site_intel_dev_scan — Full/incremental per-route scan with diffs - site_intel_quick_check — Fast static testid coverage (no Playwright) - site_intel_dev_state — Read current state without scanning - Updated /site-intel slash command with dev-scan actions - New /site-intel-dashboard dedicated slash command

#### 📦 Dev-Scan Module (8 Files, ~78KB) Complete scanning subsystem: - config-reader.js — Project config loading (226 lines) - route-catalog.js — Route discovery + reverse map (246 lines) - state-manager.js — Persistent state + history (305 lines) - git-diff.js — Changed file detection + route mapping (184 lines) - testid-checker.js — Static analysis, 11 patterns (326 lines) - scanner.js — Playwright-based scanning (361 lines) - diff-reporter.js — Categorized diff output (201 lines)

$3

`bash

Update to latest

npm update -g claude-cli-advanced-starter-pack

Sync all new features to your project

ccasp init
`

---

The Problem

Every new project means:
- ❌ Manually creating .claude/ folders and configurations
- ❌ Copy-pasting commands, hooks, and agents from old projects
- ❌ Forgetting critical settings and enforcement hooks
- ❌ Agents that don't understand your tech stack
- ❌ Hours of setup before you can actually build

The Solution

One command. Fully configured. Stack-aware.

`bash
npx claude-cli-advanced-starter-pack init
`

CCASP scans your project, detects your stack (React? FastAPI? Prisma? Playwright?), and deploys a complete .claude/ folder with:

- ✅ 64 slash commands — deployment, testing, refactoring, GitHub sync, UI testing
- ✅ 44 enforcement hooks — validation, progress tracking, agent orchestration
- ✅ Stack-specific agents — specialists that know your frameworks
- ✅ Smart updates — add features without losing customizations

---

What You Get

| Component | Count | Examples |
|-----------|-------|----------|
| Slash Commands | 64 | /deploy-full, /pr-merge, /ralph, /ui-test, /vision-list |
| Enforcement Hooks | 44 | Token tracking, agent orchestration, refactor safety, UI test auto-run |
| Tech Stack Detection | 55+ | React, Vue, FastAPI, Express, Prisma, Playwright |
| Agent Templates | L1/L2/L3 | Orchestrators → Specialists → Workers + UI testing agents |
| MCP Integrations | 50+ | Railway, Cloudflare, GitHub, Playwright |
| Skills (RAG) | 6 | agent-creator, hook-creator, refactor patterns, nvim-ui-test |

---

AI Constitution Framework

Enforce project-specific coding standards and architecture patterns with automatic validation.

$3

`bash
ccasp constitution-init --preset senior # 5% sampling, all sections
ccasp constitution-init --preset minimal # 2% sampling, security only
ccasp constitution-init --preset strict # 15% sampling, maximum coverage
`

$3

- YAML-based rules at .claude/config/constitution.yaml
- Hook enforcement with configurable sampling (1-in-20 default)
- Sensitive pattern bypass - security rules always checked
- Stop → Explain → Suggest → Refactor violation flow
- Neovim integration - edit constitution from sidebar

$3

| Section | Examples |
|---------|----------|
| code_style | Naming, types, imports |
| architecture | Patterns, boundaries |
| security | Credentials, validation |
| performance | Async, memoization |
| git | Commits, branches |
| dependencies | Versions, security |

See /ai-constitution-framework in Claude Code for full documentation.

---

See It In Action

📸 Screenshots Gallery (click to expand/collapse)

$3

/menu — Interactive Navigation Full-featured menu with all options

Project Configuration Settings GitHub, deployment, tunnels, tokens

$3

/menu — Mobile Navigation Single-key navigation, 40-char width	Setup Wizard — Feature Selection Choose commands, agents, hooks, skills
GitHub Project Setup One-click repo and Project Board sync	/claude-audit — Analysis Report CLAUDE.md analysis with score
Audit Score Breakdown Detailed scoring and summary	Actionable Recommendations Priority-ordered improvements

---

Highlight Features

$3 Auto-generates agents that know your tech: - React + Zustand + Tailwind specialist - FastAPI + SQLAlchemy backend agent - Playwright E2E testing expert - PostgreSQL + Prisma database agent	$3 Continuous test-fix cycle until green: `bash /ralph --watch ` Runs tests → Parses failures → Fixes code → Repeats. Smart Recovery: Every 3rd failed attempt deploys a web search agent to find best-practice solutions for stubborn bugs or implementation issues. Max 10 iterations. Stops on 3x same failure.
$3 Coordinate full-stack, multi-feature development: `bash /create-roadmap "Q1 Feature Sprint" ` - Break epics into phased milestones - GitHub issue tracking per phase - Customizable Project Board sync - Dependency graphs with auto-ordering - Progress dashboard across all phases	$3 Full-stack deploy to Railway + Cloudflare: `bash /deploy-full ` - Backend and frontend in parallel - Pre-flight validation - Automatic rollback on failure
$3 Break complex features into phases: `bash /phase-dev-plan "Add user authentication" ` - 95%+ success probability planning - PROGRESS.json state tracking - GitHub issue sync - L1/L2/L3 agent orchestration	$3 9-phase merge workflow with safety: `bash /pr-merge ` - Auto-stash uncommitted changes - Resolve conflicts, CI failures, reviews - Automatic rollback on any failure
$3 Multi-issue epic workflows: `bash /create-github-epic "Auth System" ` - Creates parent epic with child issues - Tracks completion across all issues - Auto-syncs with Project Board - Progress comments on epic issue	$3 Works with Happy.engineering: - 40-character max width - Card-based layouts - Single-character inputs - Auto-detected via environment
$3 Transform natural language into complete, working MVPs: `bash ccasp vision init "Build a kanban board with real-time collaboration" ` 8-Stage Orchestrated Workflow: 1. Initialize — Parse prompt, detect intent, extract features 2. Analyze — Web search for inspiration, discover npm/pip packages, match MCP servers 3. Architect — Generate Mermaid diagrams, API contracts, ASCII wireframes 4. Security — Scan dependencies with npm audit, pip-audit, OSV Scanner 5. Create Agents — Spawn specialized agents for detected tech stack 6. Execute — Autonomous development loop with self-healing 7. Validate — Run tests, verify MVP completeness 8. Complete — Final verification and checkpoint creation Features: - 🎯 Natural Language Input — "Build a todo app with React and FastAPI" - 🔍 Analysis Engine — Similar apps search, tool discovery, MCP matching - 🏗️ Architecture Planning — Mermaid diagrams, API contracts, state design - 👁️ Drift Detection — Hook-based observer with automatic plan adjustment - 🔒 Security Scanning — Pre-install vulnerability detection - 🤖 Dynamic Agents — Creates specialists based on tech stack - 🔄 Self-Healing — Automatic test failure resolution - 📊 Web Dashboard — Real-time status at http://localhost:3847 `bash Initialize and run ccasp vision init "Your app idea here" ccasp vision run my-app Multi-vision management (v2.7.0) ccasp vision list # Registry-backed listing ccasp vision cleanup # Remove orphaned visions Monitor with web dashboard ccasp vision dashboard Or use slash commands /vision-init # Decision engine auto-selects plan type /vision-status # Shows plan type indicators /vision-list # Multi-vision registry management /vision-run `

---

Architecture

After running ccasp init, your project gets:

`
.claude/
├── commands/ # 64 slash commands
│ ├── menu.md # Interactive navigation
│ ├── deploy-full.md # Full-stack deployment
│ ├── ralph.md # Test-fix loops
│ └── ...
├── agents/ # Stack-specific AI agents
│ ├── react-specialist.md
│ ├── fastapi-specialist.md
│ └── ...
├── hooks/ # 44 enforcement hooks
│ ├── ralph-loop-enforcer.js
│ ├── progress-tracker.js
│ └── ...
├── skills/ # RAG-enhanced packages
├── config/
│ └── tech-stack.json # Your detected stack
└── settings.json # Project configuration
`

Everything is customized to your package.json, config files, and directory structure.

---

Quick Start

`bash

Step 1: Install

npm install -g claude-cli-advanced-starter-pack

Step 2: Initialize (detects your stack, deploys .claude/)

ccasp init

Step 3: Restart Claude Code CLI (required for new commands)

claude .

Step 4: Use slash commands

/menu
`

$3

`bash
npx claude-cli-advanced-starter-pack init
`

---

Roadmap

$3

| Feature | Status |
|---------|--------|
| Neovim UI Testing + Appshell Layout | ✅ Complete |
| Vision Multi-Instance & Decision Engine | ✅ Complete |
| Smart Site Intel Dev-Scan | ✅ Complete |
| Vision Mode Orchestration | ✅ Complete |
| Website Intelligence (Site-Intel) | ✅ Complete |
| Anthropic MCP Registry | ✅ Complete |
| nvim-ccasp Modular Refactoring | ✅ Complete |
| JS Module Decomposition (Epic 3) | ✅ Complete |
| Jira Integration | 🔨 In Development |
| Linear Sync | 🔨 In Development |
| ClickUp Integration | 📋 Planned |
| Vision Driver Bot (VDB) v2 | 🔨 In Development |
| Multi-Repo Orchestration | 📋 Planned |

---

Version History

$3

Neovim UI Testing Framework + Appshell Layout — Feb 2026
- UI Testing Framework: Deterministic headless testing with plenary.nvim — driver (534 lines), snapshot (512 lines), 4 spec files, 3 runner scripts
- Appshell Layout: VS Code-inspired zone-based UI with icon rail, flyout, header, footer, content zone, right panel
- 6 New Commands: /ui-test, /ui-smoke, /ui-snapshot, /ui-bug, /ui-fix, /repo-guard
- 3 New Agents: ui-driver, ui-debugger, nvim-test-runner with clear delegation boundaries
- 2 New Hooks: nvim-after-edit-test (auto-smoke on Lua edit), agent-limit-compact (3 concurrent max)
- Per-Folder Documentation: 15 files across nvim-ui-testing/ subfolders (READMEs, examples, checklists)
- Integration Templates: MCP-Neovim bridge config, claude-code.nvim compatibility guide
- CLAUDE_WORKFLOW.md: Development workflow documentation for the nvim-ccasp plugin

$3

Vision Multi-Instance & Decision Engine — Feb 2026
- Vision Registry: Centralized VISION_REGISTRY.json with atomic writes, auto-rebuild from filesystem, cross-process file locking
- Decision Engine: Weighted scoring selects plan type (task-list / phase-dev-plan / roadmap / epic / vision-full) based on prompt complexity
- Slug Uniqueness: generateUniqueSlug() appends -2/-3 on collision, registry-backed enforcement
- Enhanced CLI: ccasp vision list (registry-backed with plan type column), ccasp vision cleanup (orphaned/failed removal)
- New Slash Command: /vision-list for multi-vision management
- Templates Updated: /vision-init shows decision engine result, /vision-status shows plan type indicators
- 32 New Tests: Registry, slug uniqueness, decision engine, file locking, edge cases, backward compatibility (489 total)

$3

Smart Site Intel Dev-Scan System — Feb 2026
- Dev-Scan Module: 8-file scanning subsystem (~78KB) for developer-focused per-route analysis
- Dashboard Extensions: 7 new API endpoints + Dev Scan tab with sortable route table, grade badges, sparklines, diffs
- 3 New MCP Tools: site_intel_dev_scan, site_intel_quick_check, site_intel_dev_state (12 total)
- Slash Commands: Updated /site-intel with dev-scan actions, new /site-intel-dashboard command
- Git Diff Integration: Incremental scanning based on changed files with 30% threshold for full scan promotion
- Static TestID Checker: 11 interactive element patterns, recursive 2-level import resolution, <10s for 80+ routes
- State Persistence: Per-route scores with 20-entry FIFO history, regression/improvement tracking across commits
- Barrel Exports: Full dev-scan module exports in src/site-intel/index.js

$3

Neovim Plugin Refactoring + Site Intelligence + MCP Registry — Feb 2026
- nvim-ccasp 7-Phase Refactoring: 39 files, 12,505 lines. sidebar.lua -52%, 10 new modules, health score 87/100
- Website Intelligence System: 5-layer MCP server (discovery, summarizer, graph, memory, judgment) with web dashboard
- Anthropic MCP Registry: Official API integration for 50+ verified MCP servers with auto-recommendations
- JS Module Decomposition (Epic 3): 20 files decomposed into 74 focused submodules across 3 phases
- New Hooks: agent-guardrails, file-collision-detector, quality-check, ralph-bdd-mode, tdd-enforcer
- New Skills: github-actions-generator with security references
- Test Coverage: 15+ new test files (site-intel, token-budget, roadmap-schema, delegation-protocol, etc.)

$3

Neovim Integration (nvim-ccasp) — Feb 2026
- Zero-Setup Launcher: ccasp neovim launches Neovim with CCASP pre-loaded
- Auto-Install Neovim: Detects winget/choco/scoop (Windows), brew (macOS), apt/dnf/pacman (Linux)
- Multi-Session Terminals: Up to 8 Claude CLI sessions with automatic quadrant stacking
- Color-Coded Titlebars: 8 colors (Blue, Green, Purple, Orange, Red, Cyan, Pink, Yellow) with rename/minimize/close
- Auto-Insert Mode: Click on session to start typing immediately
- Collapsible Sidebar: Mouse-clickable sections, all collapsed by default
- Prompt Injector v1.1.0: Intercept prompts, optionally enhance with GPT-5.2
- npm Package Integration: nvim-ccasp/ included in npm distribution
- Permanent Installation: ccasp nvim-setup with symlink/copy options

$3

Critical Commands & Self-Healing Updates — Feb 2026
- Critical Commands: update-check and __ccasp-sync-marker now ALWAYS update during ccasp init
- Self-Healing Updates: /update-check can now update itself, fixing the bootstrap problem
- Auto-Upgrade Detection: Missing sync marker triggers automatic ccasp init for full 11-category sync
- Backwards Compatibility: Older projects automatically upgraded to new sync system

$3

Auto-Run Init on Outdated Projects — Feb 2026
- Automatic Upgrade: /update-check now runs ccasp init when sync marker missing
- Zero-Touch Migration: Users don't need to know about the upgrade — it just works

$3

Sync Marker System — Feb 2026
- New File: __ccasp-sync-marker.md detects if project has full 11-category sync
- Version Detection: Marker indicates v2.2.16+ capabilities
- Upgrade Path: Clear messaging when older versions detected

$3

Hook File Extension Support — Feb 2026
- .cjs Support: Hooks now support .cjs extension for CommonJS compatibility
- github-progress-hook.cjs: New hook registered for TodoWrite matcher
- Settings Registration: Automatic hook registration during init

$3

Full Template Category Sync — Feb 2026
- 11 Categories: /update-check now scans ALL template directories
- New Categories: commands-dev, scripts, github/ISSUE_TEMPLATE, workflows, mcp
- Complete Coverage: No more missing files after updates

$3

Simplified Update Check — Feb 2026
- Comprehensive Sync: /update-check redesigned for complete file synchronization
- Hash Comparison: Detects customized files and skips them automatically
- Smart Merge: Preserves user modifications while adding new features

$3

Enhanced Issue Templates with Hook-Based Sync — Feb 2026
- Type-Specific Templates: 4 templates (Feature, Refactor, Bug, Testing)
- CCASP-META: Machine-parseable HTML comments for agent updates
- Generated Files Section: All issues link to source documentation
- Hook Infrastructure: PostToolUse hook detects PROGRESS.json changes
- L3 Sync Worker: Background agent updates GitHub issues automatically
- 5-Phase Implementation: Complete architecture from templates to testing

$3

Consolidated Roadmap Architecture — Feb 2026
- Unified Architecture: Task/Phase/Roadmap share same exploration file structure
- Tunnel URL Support: Development tunnel URLs integrated into deployment workflow
- Consistent Planning: All three scales generate 6-file exploration documents

$3

Menu Display Fix — Feb 2026
- ASCII Box Menu: Clean display with consistent formatting
- Display Instructions: Better guidance in /menu command

$3

Mandatory L2 Exploration Files — Feb 2026
- Enforcement: L2 specialists MUST create exploration files before implementation
- 6-File Standard: EXPLORATION_SUMMARY, CODE_SNIPPETS, REFERENCE_FILES, AGENT_DELEGATION, PHASE_BREAKDOWN, findings.json
- Quality Gates: Slash commands validate exploration docs exist

$3

Graduated Task/Phase/Roadmap Architecture — Feb 2026
- Three Planning Scales: Tasks, Phases, and Roadmaps with unified structure
- L2 Exploration System: Specialists receive full context through generated docs
- Fast Dev-Sync: Quick synchronization for development workflows
- Worktree Integration: Improved dev mode with git worktree support

$3

Dev Mode Smart Sync & Wizard UX — Feb 2026
- Smart Sync for Dev Mode: New smart-sync.js utility preserves user customizations when syncing from worktree
- Menu Worktree Sync: [W] Sync from Worktree option appears during dev mode with sync status banner
- /dev-mode-deploy-to-projects: New slash command with --dry-run, --force, --project options
- /menu-happy: Dedicated mobile menu command (40-char width) — auto-detection removed from /menu
- Wizard UX Improvements: Reinstall option, accurate template counts, better mismatch handling (#64)
- Hook Migration: Hooks migrated from .js to .cjs for CommonJS compatibility

$3

E2E Testing Integration & Happy Mode Fix — Feb 2025
- Comprehensive E2E Testing Integration: Deep Playwright integration with credential injection, tunnel services, and Ralph Loop configuration
- Happy Mode Detection Fix: BREAKING - HAPPY_SERVER_URL alone no longer triggers mobile mode; requires active session indicators
- GitHub Integration Improvements: Enhanced error handling, cleaner issue parsing, refined MCP handling
- Screenshot Gallery: Added 6 screenshots showcasing desktop and mobile UI
- Developer Guides: New wiki sections for forking, local development, contributing, and npm updates

$3

Security & Packaging — Feb 2025
- Removed hardcoded paths from npm package
- Improved template portability

$3

Vision Driver Bot & GitHub Epics — Jan 2025
- Vision Driver Bot (VDB): Autonomous development with lint fixes
- GitHub Epic System: Multi-issue epic workflows with /create-github-epic
- Init for New Projects: /init-ccasp-new-project for Happy users
- Modular Commands: 7 large files refactored for maintainability

Full v2.2.0 Release Notes →

View All Releases →

---

Top Features

$3

Autonomous MVP development from natural language:

`bash
/vision-init "Build a task management app with drag-and-drop"
`

Vision Mode is an L0+ orchestration layer above Epics, enabling:
- Natural Language → MVP: Describe your product, Claude builds it autonomously
- 8-Phase Architecture: Schema → Analysis → Architecture → Observer → Agent Factory → Security → Autonomous Execution → Integration
- Drift Detection: Observer hook monitors PROGRESS.json and ROADMAP.json for alignment
- Self-Healing: Autonomous loop retries failed tasks with intelligent fixes

`
Vision Hierarchy:
VISION (L0+) ← Natural language prompt
├── EPIC (L0) ← Generated epic with milestones
│ ├── ROADMAP (L1) ← Multi-phase roadmap
│ │ └── PHASE-DEV (L2) ← Individual phases
│ │ └── TASKS (L3) ← Atomic work units
`

Commands:
- /vision-init — Initialize from prompt (decision engine auto-selects plan type)
- /vision-status — View progress, alignment, and plan type
- /vision-list — Multi-vision registry management
- /vision-adjust — Adjust when drift detected

$3

Autonomous development workflows with automatic lint fixes:

`
VDB State (.claude/vdb/state.json)
├── Current task tracking
├── Lint error queue
├── Fix history
└── Session metrics

Workflow: Detect lint errors → Queue fixes → Apply → Verify
`

$3

Multi-issue epic workflows:

`bash
/create-github-epic "User Authentication System"
`
- Creates parent epic issue
- Links child issues automatically
- Tracks completion across issues
- Syncs with Project Board

$3

Hierarchical agent execution:

`
L1 Orchestrator (You / Main Conversation)
├── L2 Frontend Specialist (Sonnet)
│ ├── L3 Component Search Worker (Haiku)
│ └── L3 Style Analyzer Worker (Haiku)
├── L2 Backend Specialist (Sonnet)
│ └── L3 API Discovery Worker (Haiku)
├── L2 Testing Specialist (Sonnet)
└── L2 Deployment Specialist (Sonnet)
`

$3

`bash

Continuous test-fix until green

/ralph --watch

Characterization tests before refactoring

/golden-master src/utils/calculate.ts

Guided 8-step refactoring workflow

/refactor-workflow
`

$3

Zero-Setup Launch:
`bash
ccasp neovim # Launch with CCASP pre-loaded
ccasp neovim /path/to/dir # Launch in specific directory
ccasp nvim-setup # Install permanently to Neovim config
`

Features:
- Multi-Session Terminals: Up to 8 Claude CLI sessions with quadrant stacking
- Appshell Layout: VS Code-inspired zone-based UI (icon rail, flyout, header, footer, content, right panel)
- UI Testing Framework: Headless tests with plenary.nvim (/ui-test, /ui-smoke, /ui-snapshot)
- Color-Coded Titlebars: 8 colors with rename, minimize, close buttons
- Collapsible Sidebar: Mouse-clickable sections, all collapsed by default
- Control Panel: Toggle features, change models, restart agents
- Telescope Integration: Browse commands, skills, hooks with fuzzy finding

Keybindings:
| Key | Action |
|-----|--------|
| Ctrl+B | Toggle sidebar |
| Ctrl+Shift+N | New Claude session |
| ` (backtick) | Quick toggle sessions |
| Tab | Cycle sessions |
| ,p | Control Panel |
| ,g | Agent Grid |
| ,i | Toggle Prompt Injector |

lazy.nvim Config:
`lua
{
dir = "path/to/nvim-ccasp", -- Auto-generated by ccasp neovim
config = function()
require("ccasp").setup({
layout = "classic",
sidebar = { width = 40, auto_open = true },
terminal = { shell = "claude", size = 60 },
})
end,
}
`

Auto-installs Neovim on first run (winget/brew/apt detected automatically).

$3

Auto-detected. Mobile-optimized formatting:

`
┌────────────────────────────────────┐
│ [1] Issue #42 │
│ Add JWT authentication │
├────────────────────────────────────┤
│ Status: Ready │
│ Priority: High │
│ @johndoe • 2h ago │
└────────────────────────────────────┘
`

$3

`bash
/pr-merge # Current branch
/pr-merge 123 # Specific PR
/pr-merge --dry-run
`

9 phases: Identify → Checkpoint → Detect blockers → Resolve → Message → Merge method → Execute → Cleanup → Summary

$3

`bash
ccasp create-roadmap # Mode A: Natural language → phases
ccasp roadmap import # Mode B: GitHub issues → phases

/roadmap-status # Progress dashboard
/roadmap-edit # Reorder, merge, split phases
/roadmap-track # Execute with dependency checking
`

---

Commands Reference

$3

| Command | Description |
|---------|-------------|
| ccasp wizard | Vibe-friendly setup wizard |
| ccasp init | Deploy to project |
| ccasp detect-stack | Auto-detect tech stack |
| ccasp create-agent | Create L1/L2/L3 agents |
| ccasp create-hook | Create enforcement hooks |
| ccasp explore-mcp | MCP server discovery |
| ccasp neovim | Launch Neovim with CCASP (zero-setup) |
| ccasp nvim-setup | Install nvim-ccasp permanently |

$3

| Command | Description |
|---------|-------------|
| /menu | Interactive navigation |
| /deploy-full | Full-stack deployment |
| /pr-merge | Safe PR merge workflow |
| /ralph | Test-fix loops |
| /phase-dev-plan | Phased development |
| /create-github-epic | Epic workflows |
| /refactor-workflow | Guided refactoring |
| /golden-master | Characterization tests |
| /vision-init | Autonomous MVP from prompt |
| /vision-status | Vision progress dashboard |
| /vision-adjust | Adjust Vision plan |
| /vision-list | Multi-vision registry management |
| /ui-test | Neovim UI full test suite |
| /ui-smoke | Quick Neovim smoke test |
| /ui-snapshot | Layout snapshot + diff |
| /ui-fix | Guided UI debug/fix workflow |
| /repo-guard | Pre-commit quality gate |
| /site-intel | Website intelligence + dev scanning |
| /site-intel-dashboard | Dev scan dashboard viewer |

Full Commands Reference →

---

Hook Templates (44)

| Category | Hooks |
|----------|-------|
| Token & Session | token-usage-monitor, session-id-generator, context-guardian |
| Deployment | branch-merge-checker, deployment-orchestrator |
| Refactoring | ralph-loop-enforcer, refactor-verify, refactor-transaction |
| Agent Orchestration | hierarchy-validator, progress-tracker, l2-completion-reporter |
| Neovim UI Testing | nvim-after-edit-test, agent-limit-compact |
| Happy.engineering | happy-mode-detector, happy-checkpoint-manager |

---

Tech Stack Detection

Detects 55+ frameworks by reading project files (no AI):

| Category | Detected |
|----------|----------|
| Frontend | React, Vue, Angular, Svelte, Next.js, Nuxt, Astro |
| Backend | FastAPI, Express, NestJS, Django, Flask, Rails |
| Database | PostgreSQL, MySQL, MongoDB, SQLite, Redis |
| ORM | Prisma, TypeORM, SQLAlchemy, Drizzle |
| Testing | Jest, Vitest, Playwright, Cypress, pytest |
| Deployment | Railway, Vercel, Netlify, Cloudflare |

---

MCP Server Integration

`bash
ccasp explore-mcp # Interactive menu
ccasp explore-mcp --recommend # Auto-recommend for your stack
`

| Category | Servers |
|----------|---------|
| Testing | Playwright, Puppeteer, Skyvern |
| Deployment | Railway, Cloudflare, Vercel |
| Database | PostgreSQL, Supabase, Redis |
| Tunnel | ngrok, Cloudflare Tunnel |

---

Contributing

`bash
git clone https://github.com/evan043/claude-cli-advanced-starter-pack.git
cd claude-cli-advanced-starter-pack
npm install
npm run lint
npm test
``

Contributing Guidelines →

---

License

MIT © evan043

---