luna-agents 2.0.1

package/commands/ll-browser-test.md

@@ -0,0 +1,203 @@
+---
+name: ll-browser-test
+displayName: Luna Browser Test
+description: Full browser integration testing — launch app, navigate flows, screenshot, auto-fix failures
+version: 1.0.0
+category: testing
+agent: luna-full-test
+parameters:
+  - name: url
+    type: string
+    description: App URL to test (e.g., http://localhost:3000 or https://app.example.com)
+    required: true
+    prompt: true
+  - name: flow
+    type: string
+    description: Flow to test (full, auth, dashboard, billing, settings, custom, all)
+    required: false
+    prompt: true
+workflow:
+  - detect_app_and_routes
+  - launch_browser
+  - execute_test_flows
+  - capture_screenshots
+  - analyze_failures
+  - auto_fix_code
+  - retest_fixes
+  - generate_browser_report
+output:
+  - .luna/{current-project}/browser-test/
+  - .luna/{current-project}/browser-test/report.md
+prerequisites: []
+---
+
+# Luna Browser Test
+
+Launch your app in a real browser, test every flow, screenshot every page, and auto-fix failures.
+
+## What This Command Does
+
+1. **Detect** — scans your codebase for routes, pages, components
+2. **Launch** — starts Playwright browser (chromium, firefox, webkit)
+3. **Navigate** — walks through every user flow (with and without auth)
+4. **Screenshot** — captures every page state (desktop + mobile)
+5. **Analyze** — uses Vision AI to detect UI issues in screenshots
+6. **Auto-Fix** — fixes broken code, reruns the test
+7. **Loop** — keeps fixing and retesting until all flows pass
+8. **Report** — full visual report with before/after screenshots
+
+## Test Flows
+
+### Without Login (Public)
+| Flow | What It Tests |
+|------|--------------|
+| Landing Page | Hero loads, CTAs visible, responsive layout |
+| Marketing Pages | Features, pricing, about, contact |
+| Documentation | Docs pages render, navigation works, code blocks display |
+| Auth Pages | Sign-in form renders, OAuth buttons visible, validation works |
+| 404/Error Pages | Error pages render correctly |
+| SEO | Meta tags, OG images, canonical URLs, robots.txt |
+| Performance | LCP < 2.5s, CLS < 0.1, no layout shifts |
+
+### With Login (Authenticated)
+| Flow | What It Tests |
+|------|--------------|
+| Sign Up | Register with email, verify email, complete onboarding |
+| Sign In | Email/password, OAuth (Google, GitHub, Microsoft, LinkedIn) |
+| Dashboard | Data loads, widgets render, navigation works |
+| CRUD Operations | Create, read, update, delete entities |
+| Team Management | Invite member, accept invite, change role, remove member |
+| Billing | View plan, upgrade, manage payment, view invoices |
+| Settings | Update profile, change password, manage notifications |
+| Workflows | Create workflow, add steps, test run, view logs |
+| API Keys | Generate key, copy, revoke, test permissions |
+| Search | Search queries return results, filters work |
+
+### Cross-Cutting
+| Flow | What It Tests |
+|------|--------------|
+| Responsive | Every page at 375px, 768px, 1024px, 1440px |
+| Dark Mode | Toggle dark mode, all components render correctly |
+| Keyboard Navigation | Tab through all interactive elements, focus visible |
+| Error States | Empty states, loading states, error boundaries |
+| Network Errors | Offline mode, slow connection, API failures |
+| Session Expiry | Token expiry, refresh flow, redirect to login |
+
+## Usage
+
+```
+/browser-test http://localhost:3000                    # Test everything
+/browser-test http://localhost:3000 auth               # Auth flows only
+/browser-test http://localhost:3000 dashboard           # Dashboard only
+/browser-test http://localhost:3000 billing             # Billing only
+/browser-test https://staging.myapp.com all             # Test staging
+/browser-test http://localhost:3000 full                # All flows + auto-fix
+```
+
+## Auto-Fix Loop
+
+When a test fails:
+
+```
+Test fails → Screenshot failure state
+  → Vision AI analyzes screenshot
+  → Identifies the UI issue (broken layout, missing element, wrong text)
+  → Reads the source code for that component
+  → Generates fix (max 100 lines per file)
+  → Applies fix
+  → Retests the same flow
+  → If pass → move to next flow
+  → If fail → try again (max 3 attempts)
+  → If still fails → log as blocker, continue to next flow
+```
+
+## Screenshot Strategy
+
+Every page gets captured at 4 viewports:
+
+| Viewport | Width | Device |
+|----------|-------|--------|
+| Mobile | 375px | iPhone 15 |
+| Tablet | 768px | iPad |
+| Laptop | 1024px | MacBook Air |
+| Desktop | 1440px | External display |
+
+Screenshots saved to:
+```
+.luna/{project}/browser-test/screenshots/
+  landing/
+    desktop.png
+    laptop.png
+    tablet.png
+    mobile.png
+    dark-desktop.png
+    dark-mobile.png
+  dashboard/
+    desktop.png
+    mobile.png
+  auth/
+    signin-desktop.png
+    signin-mobile.png
+    signup-desktop.png
+  ...
+```
+
+## Test Credentials
+
+Provide test credentials in `.luna/rules.yaml`:
+```yaml
+browser_test:
+  url: http://localhost:3000
+  credentials:
+    email: test@example.com
+    password: TestPassword123!
+  oauth_mock: true                 # Mock OAuth providers in test
+  seed_data: true                  # Seed test data before running
+```
+
+Or pass inline:
+```
+/browser-test http://localhost:3000 --email test@example.com --password Test123!
+```
+
+## Vision AI Analysis
+
+Each screenshot is analyzed for:
+- **Layout issues** — overlapping elements, broken grids, overflow
+- **Missing content** — empty containers, broken images, missing text
+- **Style problems** — wrong colors, inconsistent spacing, font issues
+- **Accessibility** — contrast ratio, missing focus indicators, touch targets
+- **Apple HIG compliance** — spacing, typography, visual hierarchy
+- **Brand consistency** — colors match brand guide, fonts correct
+- **Responsive issues** — content cut off, horizontal scroll, tiny text
+- **Dark mode issues** — unreadable text, wrong backgrounds, missing styles
+
+## In Pipes
+
+```
+# Test then ship
+/pipe browser-test http://localhost:3000 ?>> ship !>> fix
+
+# Full flow: implement, test in browser, fix, ship
+/pipe go *5 >> browser-test http://localhost:3000 full ?>> pr !>> (fix >> browser-test) *3?
+
+# Test all viewports after UI changes
+/pipe hig >> browser-test http://localhost:3000 >> a11y >> pr
+
+# Brand check: generate brand, apply HIG, verify in browser
+/pipe brand >> hig >> browser-test http://localhost:3000 >> approve "Looks good?" >> ship
+```
+
+## Output
+
+```
+.luna/{project}/browser-test/
+  report.md                  # Full test report with pass/fail per flow
+  screenshots/               # All captured screenshots
+    {page}/{viewport}.png
+  fixes/                     # Auto-fix diffs
+    {component}-fix.diff
+  failures/                  # Failed test screenshots
+    {flow}-failure.png
+  coverage.json              # Route coverage (which pages were tested)
+```

package/commands/ll-challenge.md

@@ -0,0 +1,124 @@
+---
+name: ll-challenge
+displayName: Luna Challenge
+description: Timed coding challenges with AI judging — earn credits, climb leaderboards
+version: 1.0.0
+category: community
+agent: luna-challenge-runner
+parameters:
+  - name: difficulty
+    type: string
+    description: "Challenge difficulty level"
+    required: false
+    default: medium
+    enum: [easy, medium, hard]
+  - name: topic
+    type: string
+    description: "Optional topic filter: algorithms, api-design, refactoring, testing, security, performance"
+    required: false
+workflow:
+  - fetch_challenge
+  - start_timer
+  - monitor_submission
+  - ai_judge_evaluate
+  - calculate_score
+  - update_leaderboard
+output: []
+prerequisites: []
+---
+
+# Luna Challenge
+
+Timed coding challenges with AI judging, inspired by flow-nexus. Solve real-world coding problems, get evaluated on correctness, performance, code quality, and innovation.
+
+## What This Command Does
+
+1. **Fetch Challenge** — pulls a challenge matching your difficulty and topic preferences
+2. **Start Timer** — begins the countdown (easy: 15min, medium: 30min, hard: 60min)
+3. **Monitor Submission** — watches for your solution file
+4. **AI Judge** — evaluates your submission across four dimensions
+5. **Calculate Score** — produces a composite score out of 100
+6. **Update Leaderboard** — posts your result to the global leaderboard
+
+## Usage
+
+```
+/challenge
+/challenge --difficulty easy
+/challenge --difficulty hard --topic security
+/challenge --topic algorithms
+```
+
+## Difficulty Levels
+
+| Level | Time Limit | Complexity | Credits Earned |
+|-------|-----------|------------|----------------|
+| **Easy** | 15 minutes | Single function, clear spec | 10 credits |
+| **Medium** | 30 minutes | Multi-file, edge cases | 25 credits |
+| **Hard** | 60 minutes | Architecture, optimization | 50 credits |
+
+## Topics
+
+| Topic | Example Challenges |
+|-------|-------------------|
+| **algorithms** | Implement a priority queue, graph traversal |
+| **api-design** | Design a REST API for a feature, handle pagination |
+| **refactoring** | Clean up legacy code, extract patterns |
+| **testing** | Write tests for untested code, achieve coverage |
+| **security** | Fix vulnerabilities, implement auth flow |
+| **performance** | Optimize slow queries, reduce bundle size |
+
+## AI Judge Scoring
+
+Your submission is evaluated on four dimensions:
+
+| Dimension | Weight | What's Evaluated |
+|-----------|--------|-----------------|
+| **Correctness** | 40% | All test cases pass, edge cases handled |
+| **Performance** | 20% | Time/space complexity, no unnecessary work |
+| **Code Quality** | 25% | Readability, naming, structure, types |
+| **Innovation** | 15% | Creative approach, elegant solution |
+
+## Scoring Breakdown
+
+```
+Score = (correctness * 0.4) + (performance * 0.2) 
+      + (quality * 0.25) + (innovation * 0.15)
+
+Ranks:
+  90-100: Legendary
+  75-89:  Expert
+  60-74:  Skilled
+  40-59:  Apprentice
+  0-39:   Novice
+```
+
+## Leaderboard
+
+- **Global** — all-time top scorers across all topics
+- **Weekly** — resets every Monday, top 10 earn bonus credits
+- **Topic** — per-topic rankings for specialization bragging rights
+- **Team** — aggregate scores for org/team competitions
+
+## Challenge Flow
+
+```
+/challenge --difficulty medium --topic testing
+  > Challenge: "Write tests for UserService with 90%+ coverage"
+  > Time limit: 30 minutes
+  > Files created: .luna/challenges/challenge-2026-04-07/
+  > Timer started...
+
+[You write your solution]
+
+  > Submission detected!
+  > Judging...
+  > Score: 82/100 (Expert)
+  >   Correctness: 95/100
+  >   Performance: 70/100
+  >   Code Quality: 85/100
+  >   Innovation: 65/100
+  > Credits earned: 25
+  > Leaderboard rank: #47 globally
+```
+

package/commands/ll-changelog.md

@@ -0,0 +1,80 @@
+---
+name: ll-changelog
+displayName: Luna Changelog
+description: Auto-generate changelog from git history with semantic versioning
+version: 1.0.0
+category: documentation
+agent: luna-documentation
+parameters:
+  - name: version
+    type: string
+    description: Version number (e.g., 1.2.0) or 'auto' for semantic detection
+    required: false
+    prompt: true
+  - name: scope
+    type: string
+    description: Project or feature scope
+    required: true
+    prompt: true
+workflow:
+  - read_git_history
+  - categorize_commits
+  - determine_version
+  - generate_changelog
+  - update_package_version
+  - generate_release_notes
+output:
+  - CHANGELOG.md
+  - .luna/{current-project}/release-notes.md
+prerequisites: []
+---
+
+# Luna Changelog
+
+Auto-generate changelogs from git commits with semantic versioning.
+
+## What This Command Does
+
+1. **Read** — parses git log since last tag/release
+2. **Categorize** — groups commits into Features, Fixes, Breaking Changes, etc.
+3. **Version** — determines semver bump (major/minor/patch) from commit types
+4. **Generate** — creates/updates CHANGELOG.md with new section
+5. **Update** — bumps version in package.json
+6. **Release Notes** — generates GitHub-ready release notes
+
+## Commit Categories
+
+| Prefix | Category | Version Bump |
+|--------|----------|-------------|
+| `feat:` | Features | minor |
+| `fix:` | Bug Fixes | patch |
+| `BREAKING:` | Breaking Changes | major |
+| `perf:` | Performance | patch |
+| `docs:` | Documentation | — |
+| `refactor:` | Code Changes | — |
+| `test:` | Tests | — |
+| `chore:` | Maintenance | — |
+
+## Usage
+
+```
+/changelog              # Auto-detect version
+/changelog 2.0.0        # Specific version
+/changelog auto         # Semantic auto-bump
+```
+
+## Output Format
+
+```markdown
+## [1.2.0] - 2026-03-29
+
+### Features
+- Add team billing page (#42)
+- Support LinkedIn OAuth provider (#38)
+
+### Bug Fixes
+- Fix password reset 401 error (#41)
+
+### Breaking Changes
+- Remove legacy auth middleware (#40)
+```

package/commands/ll-ci.md

@@ -0,0 +1,78 @@
+---
+name: ll-ci
+displayName: Luna CI Pipeline
+description: Generate CI/CD pipeline — GitHub Actions or GitLab CI with lint, test, build, deploy stages
+version: 1.0.0
+category: devops
+agent: luna-deployment
+parameters:
+  - name: platform
+    type: string
+    description: CI platform (github-actions, gitlab-ci, bitbucket)
+    required: false
+    prompt: true
+  - name: scope
+    type: string
+    description: Project or feature scope
+    required: true
+    prompt: true
+workflow:
+  - detect_project_stack
+  - generate_ci_config
+  - add_quality_gates
+  - add_deploy_stages
+  - add_notifications
+  - validate_config
+  - generate_ci_report
+output:
+  - .github/workflows/ or .gitlab-ci.yml
+  - .luna/{current-project}/ci-report.md
+prerequisites: []
+---
+
+# Luna CI Pipeline
+
+Generate production-grade CI/CD pipelines tailored to your stack.
+
+## What This Command Does
+
+1. **Detect** — identifies framework, package manager, test runner, deploy target
+2. **Generate** — creates CI config with all stages
+3. **Quality Gates** — adds lint, type-check, test, coverage, security scan
+4. **Deploy** — staging + production deploy with approval gates
+5. **Notifications** — Slack/Discord/email on failure
+6. **Validate** — checks config syntax
+7. **Report** — documents pipeline stages and triggers
+
+## Pipeline Stages
+
+```
+PR Opened:
+  ├── Install dependencies (cached)
+  ├── Lint + Format check
+  ├── Type check (tsc --noEmit)
+  ├── Unit tests + coverage (>=90%)
+  ├── Security scan (npm audit)
+  └── Build validation
+
+Merge to main:
+  ├── All PR checks
+  ├── Integration tests
+  ├── E2E tests (Playwright)
+  ├── Deploy to staging
+  ├── Smoke tests on staging
+  └── Deploy to production (manual approval)
+
+Scheduled (nightly):
+  ├── Full E2E suite
+  ├── Dependency vulnerability scan
+  └── Performance benchmark
+```
+
+## Usage
+
+```
+/ci                          # Auto-detect platform
+/ci github-actions           # Specific platform
+/ci gitlab-ci                # GitLab
+```

package/commands/ll-claude-instructions.md

@@ -0,0 +1,80 @@
+---
+name: ll-claude-instructions
+displayName: Luna Claude Instructions Builder
+description: Generate a comprehensive instructions file for the Claude Chrome browser extension from your codebase
+version: 1.0.0
+category: documentation
+agent: luna-claude-instructions
+parameters:
+  - name: scope
+    type: string
+    description: Project or feature scope for instructions generation
+    required: true
+    prompt: true
+workflow:
+  - scan_all_pages_and_routes
+  - inventory_menus_buttons_forms
+  - map_user_flows
+  - catalogue_features_and_interactions
+  - generate_claude_instructions_file
+output:
+  - .luna/{current-project}/claude-instructions.md
+prerequisites:
+  - source_code
+---
+
+# Luna Claude Instructions Builder
+
+Generates a comprehensive instructions file for the Claude Chrome browser extension by analysing your entire codebase. The output covers all pages, menus, buttons, flows, routing, and features — ready to paste into the extension.
+
+## What This Command Does
+
+This command crawls your project, inventories every UI element, maps all routes and navigation, documents user flows, and produces a single markdown file optimised for the Claude browser extension's custom instructions field.
+
+## Prerequisites
+
+Requires in your current project:
+- Source code with page/route components
+
+No prior Luna artifacts are required — this skill works directly from code.
+
+## Usage Instructions
+
+When you run this command, you'll be prompted for the scope:
+- Press **ENTER** for full-project instructions
+- Type **feature-name** for feature-specific instructions
+
+## Execution Steps
+
+1. **Page & Route Scan**: Discovers all pages, routes, and URL patterns
+2. **UI Inventory**: Catalogues menus, buttons, forms, modals, and interactive elements
+3. **Flow Mapping**: Traces user flows from entry to completion
+4. **Feature Catalogue**: Groups features with their UI elements and API calls
+5. **Instructions Generation**: Produces `claude-instructions.md` for the extension
+
+## Output Files
+
+Creates in your current project:
+- `.luna/{current-project}/claude-instructions.md`
+
+The file includes:
+- Application overview and tech stack
+- Complete page inventory with purpose and UI elements
+- Navigation map (sidebar, header, routing hierarchy)
+- Feature catalogue with buttons, forms, and actions
+- User flow descriptions step-by-step
+- Component library reference
+- Keyboard shortcuts
+
+## Next Steps in Workflow
+
+After generating instructions:
+- Paste into the Claude Chrome extension custom instructions
+- Run `/luna-routemap` for a dedicated route-only map
+- Run `/luna-flowdocs` for detailed Mermaid flow diagrams
+
+## Tips
+
+- Run this after every major UI change to keep instructions current
+- Combine with `/luna-codemap` for a comprehensive developer context
+- The output is optimised for AI context — concise but complete

package/commands/ll-clone.md

@@ -0,0 +1,82 @@
+---
+name: ll-clone
+displayName: Luna Clone
+description: Analyze any app/website and generate a complete clone with your own design system and improvements
+version: 1.0.0
+category: creation
+agent: luna-design-architect
+parameters:
+  - name: url
+    type: string
+    description: URL of the app to analyze and clone
+    required: true
+    prompt: true
+  - name: improvements
+    type: string
+    description: What to improve over the original (optional)
+    required: false
+mcp_servers:
+  - playwright
+  - puppeteer
+  - fetch
+  - zai-mcp-server
+  - image-extractor
+  - accessibility-scanner
+  - sequential-thinking
+  - ruflo
+---
+
+# /clone �� Reverse Engineer Any App
+
+Point at any live app. Luna screenshots every page, extracts the design system, maps the architecture, and generates a complete codebase with your improvements baked in.
+
+## Pipeline
+
+```
+/clone https://linear.app --improvements "add AI triage, better mobile UX"
+                    │
+    ┌───────────────┼───────────────┐
+    ▼               ▼               ▼
+ CAPTURE         EXTRACT         ANALYZE
+ playwright      zai-mcp         fetch
+ screenshots     UI-to-code      API routes
+ all pages       design tokens   data model
+ all viewports   components      auth flow
+    │               │               │
+    └───────────────┼───────────────┘
+                    ▼
+              SYNTHESIZE
+              • Design system (colors, fonts, spacing)
+              • Component library (React/Vue/Svelte)
+              • Page layouts (responsive)
+              • API schema (OpenAPI)
+              • Data model (Prisma/Drizzle)
+              • Auth flow (NextAuth/Clerk)
+              • Your improvements layered on top
+                    │
+                    ▼
+              GENERATE
+              • Full project scaffold
+              • Every component implemented
+              • Every page routed
+              • API stubs with types
+              • Tests for critical paths
+              • CI/CD pipeline
+              • Deployment config
+```
+
+## Usage
+
+```bash
+/clone https://notion.so                          # Full clone
+/clone https://stripe.com/dashboard --improvements "simpler onboarding"
+/clone https://vercel.com --improvements "add team chat, mobile app"
+/clone ./competitor-screenshot.png                 # Clone from screenshot
+```
+
+## In Pipes
+
+```bash
+/pipe clone https://app.com >> hig >> brand >> go *10 >> browser-test >> ship
+/pipe clone https://app.com >> improve "add AI features" >> test >> deploy
+```