luna-agents 2.0.1

package/commands/ll-autopilot.md

@@ -0,0 +1,113 @@
+---
+name: ll-autopilot
+displayName: Luna Autopilot
+description: Fully autonomous development — give Luna a goal, it plans, codes, tests, reviews, fixes, and ships without intervention
+version: 1.0.0
+category: orchestration
+agent: luna-task-executor
+parameters:
+  - name: goal
+    type: string
+    description: What you want built, fixed, or shipped — in plain English
+    required: true
+    prompt: true
+  - name: budget
+    type: number
+    description: "Max iterations before stopping for human review (default: 20)"
+    required: false
+    default: 20
+  - name: auto_ship
+    type: boolean
+    description: "Auto-deploy when done? (default: false — stops for approval)"
+    required: false
+    default: false
+mcp_servers:
+  - ruflo
+  - git
+  - memory
+  - sequential-thinking
+  - playwright
+  - zai-mcp-server
+  - accessibility-scanner
+  - fetch
+---
+
+# /autopilot — Set the Goal, Walk Away
+
+The most powerful command in Luna. Give it a goal in plain English. Luna autonomously plans, implements, tests, reviews, fixes, and optionally ships — learning from each iteration.
+
+## Autonomous Loop
+
+```
+/autopilot "Add Stripe billing with free trial, pro plan, and team plan"
+                    │
+                    ▼
+         ┌─── THINK ──────────┐
+         │  Analyze the goal    │
+         │  Break into tasks    │
+         │  Identify risks      │
+         │  Plan approach       │
+         │  (sequential-thinking)│
+         └────────┬────────────┘
+                  │
+         ┌─── EXECUTE ────────┐
+         │  For each task:      │
+         │  ├── Code it         │
+         │  ├── Test it         │
+         │  ├── Review it       │
+         │  ├── Fix issues      │
+         │  └── Mark complete   │
+         │  (ruflo swarm)       │
+         └────────┬────────────┘
+                  │
+         ┌─── VERIFY ─────────┐
+         │  Run full test suite │
+         │  Visual QA check     │
+         │  Security scan       │
+         │  A11y compliance     │
+         │  Performance check   │
+         └────────┬────────────┘
+                  │
+              PASS? ──NO──→ FIX ──→ back to EXECUTE
+                │
+               YES
+                │
+         ┌─── DELIVER ────────┐
+         │  Generate changelog  │
+         │  Create PR or deploy │
+         │  Update docs         │
+         │  Store learnings     │
+         └─────────────────────┘
+```
+
+## The Difference
+
+| Traditional | Luna Autopilot |
+|------------|---------------|
+| You plan tasks | Luna plans tasks |
+| You code each one | Luna codes each one |
+| You run tests manually | Luna runs tests automatically |
+| You fix failures | Luna fixes failures |
+| You request review | Luna reviews its own code |
+| You deploy manually | Luna deploys with confidence |
+| **Hours of work** | **One command, walk away** |
+
+## Usage
+
+```bash
+/autopilot "Add user authentication with email, Google OAuth, and magic links"
+/autopilot "Refactor the API from REST to GraphQL" --budget 30
+/autopilot "Fix all accessibility violations" --budget 15
+/autopilot "Add dark mode to the entire dashboard" --auto-ship
+/autopilot "Build a CLI that generates reports from our API"
+/autopilot "Migrate from Prisma to Drizzle ORM"
+```
+
+## Safety
+
+- **Budget limit**: Stops after N iterations (default 20) for human review
+- **Auto-ship disabled by default**: Always asks before deploying
+- **Checkpoint commits**: Every successful task gets its own git commit
+- **Rollback ready**: Can undo any autopilot session with one command
+- **Learning**: Each autopilot run makes the next one smarter
+```

package/commands/ll-boost-finsavvy.md

@@ -0,0 +1,106 @@
+---
+name: ll-boost-finsavvy
+displayName: Luna Boost FinsavvyAI Projects
+description: Analyze and boost any FinsavvyAI portfolio project using shared libraries, Claw Gateway, and cross-project synergies
+version: 1.0.0
+category: infrastructure
+agent: luna-task-executor
+parameters:
+  - name: path
+    type: string
+    description: "Project path"
+    required: false
+    default: "."
+    prompt: true
+  - name: action
+    type: string
+    description: "analyze (assess + plan), connect (register on gateway), sync (apply shared libs)"
+    required: false
+    default: analyze
+prerequisites: []
+---
+
+# Luna Boost FinsavvyAI Projects
+
+Analyze any project in the FinsavvyAI portfolio and generate a plan using shared private infrastructure, @finsavvyai libraries, and cross-project synergies.
+
+## What This Knows About
+
+### Shared Infrastructure (Deployed)
+- **Claw Gateway** — shared AI proxy at claw-gateway.workers.dev (8 projects connected)
+- **ReasoningBank** — KV prompt cache (30% token savings)
+- **Agent Booster** — skip LLM for simple transforms (<1ms, $0)
+- **Context Packing** — trim context (40-60% savings)
+- **Smart Router** — self-learning model selection
+- **Hybrid Search** — sparse+dense RRF fusion
+- **Credit System** — gamification with achievements
+- **Queen-Led Swarm** — multi-agent task decomposition (OpenSyber)
+- **Self-Learning SDK** — client-side caching + outcome tracking
+
+### @finsavvyai Shared Libraries (Master Ship Plan)
+- **@finsavvyai/pay** — Stripe + LemonSqueezy unified payments
+- **@finsavvyai/auth** — Clerk, Supabase, CF Access, JWT middleware
+- **@finsavvyai/test-config** — Vitest/Playwright presets, 95%+ coverage
+- **@finsavvyai/llm** — Multi-provider AI client with fallback chains
+- **@finsavvyai/monitor** — Sentry + Prometheus + structured logging
+- **@finsavvyai/cf-stack** — Cloudflare Workers toolkit (Hono + D1 + KV)
+- **@finsavvyai/db** — Drizzle schema templates + migration helpers
+- **@finsavvyai/ui** — Apple HIG design system components
+
+### Portfolio Projects (44 total, 8 on gateway)
+Connected: LunaOS, CodeRailFlow, Aegis, OpenSyber, Push-CI, PipeWarden, QueryFlux, Qestro
+
+### Revenue Bundles (Master Ship Plan)
+- **Security Suite** — OpenSyber + PipeWarden + Coderail.dev
+- **DevX Platform** — QueryFlux + MCPOverflow + AutoBoot
+- **FinTech Bundle** — FinTech Suite + PipeWarden
+- **Israeli Market** — YallaBye + Moneh Hacham + Hashmal
+- **AI Agents** — LunaOS + FinSavvyAI + A2A + Skill Seekers
+
+### Shipping Waves
+- Wave 1 (ship now): Luna-OS, FinSavvyAI, FinTech Suite, Skill Seekers, SubsForge, CoderailFlow, OpenSyber, AutoBoot
+- Wave 2 (quick builds): PipeWarden, TenantIQ, Qestro, DevWrapped, UPM, ViralSplit, ScanGenie
+- Wave 3-5: remaining 24 projects
+
+## Analysis Steps
+
+1. **Identify** — what is this project, what stack, what wave
+2. **Check gateway** — is it registered on Claw? If not, register it
+3. **Check shared libs** — which @finsavvyai packages does it need
+4. **Check intelligence features** — which of the 9 features apply
+5. **Check synergies** — which other projects does it complement
+6. **Check bundle** — which revenue bundle does it belong to
+7. **Generate plan** — prioritized steps with effort estimates
+
+## Usage
+
+```bash
+# Analyze current project
+/boost-finsavvy
+
+# Register on gateway + apply shared libs
+/boost-finsavvy . connect
+
+# Sync shared library versions
+/boost-finsavvy . sync
+```
+
+## Output
+
+```
+.luna/{project}/boost-finsavvy/
+  status.md             # Gateway connection, shared lib usage
+  plan.md               # Integration steps
+  synergies.md          # Cross-project opportunities
+  bundle.md             # Revenue bundle position
+```
+
+## In Pipes
+
+```bash
+# Full portfolio boost
+/pipe boost-finsavvy /path/to/project >> connect-infra . >> boost-project .
+
+# Prepare for shipping
+/pipe boost-finsavvy . >> site-audit https://myapp.com >> fix >> ship
+```

package/commands/ll-boost-org.md

@@ -0,0 +1,161 @@
+---
+name: ll-boost-org
+displayName: Luna Boost Organization
+description: Scan a GitHub org/user's repos, find cross-repo synergies, and map open-source tools to boost the entire portfolio
+version: 1.0.0
+category: analysis
+agent: luna-task-executor
+parameters:
+  - name: org
+    type: string
+    description: "GitHub org or username (e.g., 'finsavvyai', 'lunaos-ai', 'vercel')"
+    required: true
+    prompt: true
+  - name: depth
+    type: string
+    description: "quick (README only), full (+ key files), deep (+ code analysis)"
+    required: false
+    default: full
+  - name: boost
+    type: string
+    description: "What to optimize: synergies, open-source, shared-libs, all"
+    required: false
+    default: all
+prerequisites: []
+---
+
+# Luna Boost Organization
+
+Give it any GitHub org or user — it scans every repo, discovers how they relate, finds shared patterns, and maps open-source tools to supercharge the entire portfolio.
+
+## What It Does
+
+### Phase 1: Scan
+- Fetches all repos via `gh api users/{org}/repos`
+- For each repo: reads README, package.json/go.mod, key source files
+- Classifies: stack, domain, maturity, stars, activity
+
+### Phase 2: Correlate
+- Detects shared dependencies across repos
+- Identifies duplicate code patterns (same auth, billing, logging)
+- Finds repos that should share a library but don't
+- Maps which repos complement each other (e.g., API + frontend)
+- Identifies orphan repos (no relation to anything)
+
+### Phase 3: Map Open Source
+For each repo AND for cross-repo opportunities:
+
+**AI / LLM layer:**
+- [llamafile](https://github.com/mozilla-ai/llamafile) — any repo doing AI can run offline
+- [any-llm](https://github.com/mozilla-ai/any-llm) — Python repos with multiple LLM providers
+
+**Search / Data:**
+- [RuVector](https://github.com/ruvnet/RuVector) — any repo with vector search or RAG
+- Hybrid search pattern — combine keyword + semantic
+
+**Voice / Media:**
+- [Voicebox](https://github.com/jamiepine/voicebox) — any repo generating content
+- [LLaMA-Mesh](https://github.com/nv-tlabs/LLaMA-Mesh) — marketing 3D heroes
+
+**Testing / Quality:**
+- [flakestress](https://github.com/bradfitz/flakestress) — any repo with tests
+- [Perfetto](https://github.com/google/perfetto) — any web app or API
+
+**UI / Charts:**
+- [Victory](https://github.com/FormidableLabs/victory) — any repo with a dashboard
+
+**Networking:**
+- [Tailscale](https://github.com/tailscale/tailscale) — repos that talk to each other
+
+**Architecture patterns:**
+- ruflo patterns — any repo doing AI agent orchestration
+- Dossier patterns — any repo needing visual planning
+- flow-nexus patterns — any repo that could benefit from gamification
+
+### Phase 4: Shared Library Plan
+Identifies what should be extracted into shared packages:
+- Common auth patterns → `@{org}/auth`
+- Common payment logic → `@{org}/pay`
+- Common test config → `@{org}/test-config`
+- Common UI components → `@{org}/ui`
+- Common LLM client → `@{org}/llm`
+
+### Phase 5: Revenue Bundles
+Groups repos into complementary product bundles:
+- Security suite (monitoring + scanning + compliance)
+- Developer toolkit (CLI + IDE + testing)
+- AI platform (agents + orchestration + marketplace)
+
+## Usage
+
+```bash
+# Scan your own org
+/boost-org finsavvyai
+
+# Scan any GitHub org
+/boost-org vercel
+/boost-org anthropics
+
+# Scan a user
+/boost-org karpathy
+
+# Quick scan (READMEs only)
+/boost-org finsavvyai quick
+
+# Focus on open-source mapping only
+/boost-org finsavvyai full open-source
+```
+
+## Output
+
+```
+.luna/boost-org/{org}/
+  repos.md              # All repos classified
+  correlation-map.md    # How repos relate to each other
+  duplicates.md         # Shared patterns that should be libraries
+  open-source-map.md    # Which tools apply to which repos
+  shared-lib-plan.md    # Proposed shared packages
+  bundles.md            # Revenue bundle groupings
+  roadmap.md            # Prioritized cross-repo improvements
+  orphans.md            # Repos with no clear connection
+```
+
+## Example Output (for an org with 10+ repos)
+
+```markdown
+# finsavvyai — 39 Repos Analyzed
+
+## Correlation Clusters
+Cluster 1: AI Platform (5 repos)
+  luna-os ←→ finsavvyai ←→ skill-seekers ←→ a2a-framework
+  Shared: LLM calls, agent orchestration, MCP protocol
+
+Cluster 2: Security (3 repos)
+  opensyber ←→ pipewarden ←→ coderail-dev
+  Shared: security scanning, compliance, monitoring
+
+Cluster 3: Fintech (3 repos)
+  fintech-suite ←→ global-remit ←→ hashmal
+  Shared: payment processing, KYC, compliance
+
+## Shared Library Candidates
+1. LLM Client (used in 12 repos) → @finsavvyai/llm
+2. Auth Middleware (used in 9 repos) → @finsavvyai/auth
+3. Payment SDK (used in 11 repos) → @finsavvyai/pay
+
+## Open Source Boost Map
+llamafile → 8 repos (any with AI)
+Victory → 6 repos (any with dashboard)
+flakestress → all 39 repos (testing)
+Perfetto → 12 repos (web apps)
+```
+
+## In Pipes
+
+```bash
+# Scan org then boost individual projects
+/pipe boost-org finsavvyai >> boost-project /path/to/top-priority
+
+# Full portfolio optimization
+/pipe boost-org myorg >> leverage user:competitor
+```

package/commands/ll-boost-project.md

@@ -0,0 +1,118 @@
+---
+name: ll-boost-project
+displayName: Luna Boost Project
+description: Analyze any project and generate an integration plan using open-source tools from GitHub
+version: 2.0.0
+category: analysis
+agent: luna-task-executor
+parameters:
+  - name: path
+    type: string
+    description: "Project path (e.g., '.' for current)"
+    required: false
+    default: "."
+    prompt: true
+  - name: depth
+    type: string
+    description: "quick, full, deep"
+    required: false
+    default: full
+prerequisites: []
+---
+
+# Luna Boost Project — Leverage Open Source to Supercharge Any Project
+
+Scans your codebase, classifies what it does, and maps it against 20+ curated open-source tools to generate a specific integration plan.
+
+## Open-Source Tool Library
+
+### AI / LLM
+- [llamafile](https://github.com/mozilla-ai/llamafile) — Run LLMs as single executables, free, offline
+- [any-llm](https://github.com/mozilla-ai/any-llm) — Unified Python SDK for multiple LLM providers
+- [nanoGPT](https://github.com/karpathy/nanoGPT) — Minimal GPT training (~300 lines)
+- [llm.c](https://github.com/karpathy/llm.c) — Train GPT-2/3 in raw C/CUDA
+
+### 3D / Visual
+- [LLaMA-Mesh](https://github.com/nv-tlabs/LLaMA-Mesh) — Text-to-3D mesh generation
+- [3DGRUT](https://github.com/nv-tlabs/3dgrut) — 3D Gaussian ray tracing renderer
+- [PPISP](https://github.com/nv-tlabs/ppisp) — Camera correction for 3D renders
+
+### Voice
+- [Voicebox](https://github.com/jamiepine/voicebox) — Free local TTS + voice cloning, 23 languages
+
+### Performance / Testing
+- [Perfetto](https://github.com/google/perfetto) — Performance tracing + SQL analysis
+- [flakestress](https://github.com/bradfitz/flakestress) — Detect flaky tests under stress
+
+### Networking
+- [Tailscale](https://github.com/tailscale/tailscale) — Zero-config WireGuard mesh VPN
+
+### UI
+- [Victory](https://github.com/FormidableLabs/victory) — Composable React charts
+- [Spacedrive](https://github.com/spacedriveapp/spacedrive) — Rust + Tauri desktop reference
+
+### Agents
+- [Agent of Empires](https://github.com/njbrake/agent-of-empires) — Parallel AI agents with git worktrees
+
+### Search / Database
+- [RuVector](https://github.com/ruvnet/RuVector) — Self-learning vector DB, hybrid search, graph RAG
+
+### Reference Architectures
+- [ruflo](https://github.com/ruvnet/ruflo) — Self-learning agent orchestration patterns
+- [flow-nexus](https://github.com/ruvnet/flow-nexus) — Gamified agentic platform patterns
+- [Dossier](https://github.com/ruvnet/Dossier) — Visual planning + context control patterns
+
+### Community
+- [KarpathyTalk](https://github.com/karpathy/KarpathyTalk) — Dev social platform (Go + SQLite + htmx)
+- [Inbox Zero](https://github.com/elie222/inbox-zero) — AI email management
+- [GitNexus](https://github.com/abhigyanpatwari/GitNexus) — Repo analytics
+
+## Matching Logic
+
+### By Stack
+- **Go** → llamafile, Perfetto, flakestress, Tailscale
+- **TypeScript/React** → Victory, Voicebox, any-llm
+- **Python** → any-llm, llamafile, nanoGPT, llm.c
+- **Rust** → Spacedrive patterns, 3DGRUT
+
+### By Domain
+- **SaaS** → Victory, Voicebox, Inbox Zero patterns
+- **DevTool** → llamafile, Perfetto, flakestress, Agent of Empires
+- **FinTech** → RuVector hybrid search, Tailscale
+- **AI/ML** → llamafile, nanoGPT, llm.c, LLaMA-Mesh
+- **Testing** → flakestress, Perfetto
+- **Marketing** → LLaMA-Mesh, Voicebox, Victory
+
+### By Gap
+- No charts → Victory
+- No tests → flakestress
+- No voice → Voicebox
+- No offline AI → llamafile
+- No search → RuVector
+- No perf monitoring → Perfetto
+- No 3D → LLaMA-Mesh
+
+## Usage
+
+```bash
+/boost-project
+/boost-project /path/to/project
+/boost-project . deep
+```
+
+## Output
+
+```
+.luna/{project}/boost/
+  analysis.md           # Project classification
+  tool-mapping.md       # Which tools apply and why
+  integration-plan.md   # Prioritized steps with links
+  quick-wins.md         # < 1 hour improvements
+```
+
+## In Pipes
+
+```bash
+/pipe boost-project . >> go
+/pipe boost-project . >> site-audit https://myapp.com
+```

package/commands/ll-brand.md

@@ -0,0 +1,150 @@
+---
+name: ll-brand
+displayName: Luna Brand Builder
+description: Generate complete brand identity from codebase — colors, logo, vision, and style guide
+version: 1.0.0
+category: branding
+agent: luna-brand
+parameters:
+  - name: scope
+    type: string
+    description: Project or feature scope for brand generation
+    required: true
+    prompt: true
+  - name: style
+    type: string
+    description: Brand style preference (modern, minimal, playful, corporate, bold, elegant)
+    required: false
+    prompt: true
+workflow:
+  - analyze_codebase_identity
+  - extract_existing_colors_and_styles
+  - generate_brand_foundation
+  - create_color_system
+  - generate_logo_specifications
+  - create_typography_system
+  - build_visual_identity_guide
+  - generate_ai_logo_prompts
+  - create_brand_report
+output:
+  - .luna/{current-project}/brand/
+  - .luna/{current-project}/brand/brand-guide.html
+  - .luna/{current-project}/brand/brand-report.md
+prerequisites: []
+---
+
+# Luna Brand Builder
+
+Generate a complete brand identity from your codebase — analyzes your project to create colors, logo specs, typography, vision statement, and a comprehensive visual identity guide.
+
+## What This Command Does
+
+1. **Analyzes your codebase** to extract existing visual identity signals:
+   - Tailwind/CSS color tokens, design variables
+   - Existing logos, favicons, icons
+   - Product name, tagline, description from package.json/README
+   - UI patterns, component library usage
+   - Target audience from docs and marketing copy
+2. **Generates brand foundation**:
+   - Mission statement
+   - Vision statement
+   - Brand values (3-5 core values)
+   - Brand voice and tone guidelines
+   - Elevator pitch (30-second, 60-second)
+   - Tagline options (3-5 variants)
+3. **Creates complete color system**:
+   - Primary, secondary, accent colors with hex/RGB/HSL
+   - Semantic colors (success, warning, error, info)
+   - Light and dark mode palettes
+   - Contrast-checked accessibility (WCAG AA/AAA)
+   - CSS custom properties and Tailwind config
+4. **Generates logo specifications**:
+   - Logo concept description and rationale
+   - Symbol/mark guidelines
+   - Wordmark typography
+   - Logo variations (full, icon-only, horizontal, vertical)
+   - Clear space and minimum size rules
+   - Usage on light/dark backgrounds
+   - AI generation prompts for DALL-E, Midjourney, and Stable Diffusion
+5. **Creates typography system**:
+   - Primary font (headings) and secondary font (body)
+   - Font scale with rem/px values
+   - Line height and letter spacing
+   - Font weight usage guidelines
+6. **Builds visual identity guide** (standalone HTML):
+   - Interactive color swatches with copy-to-clipboard
+   - Typography specimens
+   - Logo usage examples
+   - Spacing and layout grid
+   - Component style preview
+   - Do's and don'ts section
+
+## Prerequisites
+
+No prerequisites required. This command works on any codebase.
+
+If you have existing Luna docs, they will enrich the analysis:
+- `.luna/{current-project}/requirements.md`
+- `.luna/{current-project}/design.md`
+
+## Usage Instructions
+
+When you run this command:
+- **Scope**: Press ENTER for project-level brand or type product name
+- **Style**: Choose a direction or press ENTER for auto-detect:
+  - `modern` — clean lines, gradient accents, SF Pro / Inter
+  - `minimal` — monochrome-forward, generous whitespace
+  - `playful` — rounded shapes, vibrant palette, friendly tone
+  - `corporate` — structured, trustworthy, serif + sans-serif
+  - `bold` — high contrast, large type, statement colors
+  - `elegant` — refined palette, thin weights, luxury feel
+
+## Execution Steps
+
+1. **Codebase Scan**: Reads package.json, README, CSS/Tailwind configs, existing assets
+2. **Identity Extraction**: Pulls product name, description, existing colors, fonts, logos
+3. **Brand Foundation**: Generates mission, vision, values, voice, taglines
+4. **Color System**: Creates harmonious palette from existing colors or generates new one
+5. **Logo Specs**: Defines logo concept, variations, clear space, and AI generation prompts
+6. **Typography**: Selects font pairing, defines scale, weights, and spacing
+7. **HTML Guide**: Builds interactive standalone brand guide page
+8. **AI Prompts**: Creates detailed prompts for generating logo with DALL-E/Midjourney/SD
+9. **Brand Report**: Summary document with all brand assets and usage guidelines
+
+## Output Files
+
+Creates in your current project:
+```
+.luna/{current-project}/brand/
+  brand-foundation.md          # Mission, vision, values, voice
+  color-system.md              # Full color palette with codes
+  color-tokens.css             # CSS custom properties
+  tailwind-brand.config.ts     # Tailwind color/font config
+  logo-specifications.md       # Logo concept, rules, variations
+  logo-prompts.md              # AI image generation prompts (DALL-E, MJ, SD)
+  typography-system.md         # Font pairing, scale, spacing
+  brand-guide.html             # Interactive standalone HTML brand guide
+  brand-assets/
+    color-palette.svg          # Visual color palette export
+  brand-report.md              # Complete brand summary
+```
+
+## Next Steps in Workflow
+
+After brand generation:
+```bash
+/luna-design      # Use brand tokens in technical design
+/luna-hig         # Apply Apple HIG with your brand colors
+/luna-auth        # Generate auth UI matching your brand
+/luna-docs        # Style documentation with brand identity
+```
+
+## Tips
+
+- Run early in the project lifecycle to establish visual consistency
+- The HTML brand guide is shareable — send it to designers and stakeholders
+- AI logo prompts are tuned per platform (DALL-E vs Midjourney have different syntax)
+- Color system auto-checks WCAG contrast ratios for accessibility
+- Tailwind config output can be dropped directly into your project
+- Re-run after major design changes to keep brand guide current
+- Works best when package.json has `name`, `description`, and `keywords` fields