adaptive-memory-multi-model-router 1.2.2 → 1.3.1

package/src/skills/TMLPD.md

@@ -0,0 +1,540 @@
+# TreeQuest Multi-LLM Parallel Deployment Strategy
+
+**Shortcut:** `/TMLPD`
+
+**Description:** Deploy multiple TreeQuest AI agents in parallel using multiple LLM providers for complex, multi-phase development tasks. This skill orchestrates parallel execution across different AI models to maximize throughput, leverage diverse capabilities, and accelerate project completion.
+
+---
+
+## Overview
+
+This strategy enables parallel deployment of TreeQuest agents across multiple LLM providers simultaneously, each handling specialized aspects of a project. By leveraging the strengths of different AI models (e.g., Claude for reasoning, GPT-4 for code generation, Gemini for research), we can:
+
+- **Accelerate Development:** Execute multiple independent tasks in parallel
+- **Optimize Cost-Performance:** Route tasks to the most cost-effective suitable model
+- **Leverage Specialized Capabilities:** Use models for their strengths (vision, reasoning, coding, research)
+- **Ensure Redundancy:** Run critical tasks on multiple models for verification
+
+---
+
+## Prerequisites
+
+### Required Setup
+
+1. **TreeQuest CLI Installed:**
+   ```bash
+   # Check installation
+   treequest status
+
+   # If not installed
+   pip install treequest-ai
+   ```
+
+2. **Multiple API Keys Configured:**
+   ```bash
+   # View current configuration
+   treequest config --show
+
+   # Edit configuration to add providers
+   # File: ~/.config/treequest/config.yaml
+   ```
+
+3. **Minimum Providers Recommended:**
+   - **Anthropic** (Claude Sonnet/Opus) - Complex reasoning and architecture
+   - **OpenAI** (GPT-4 Turbo) - Code generation and debugging
+   - **Google** (Gemini Pro) - Research and documentation
+   - **Groq/Cerebras** (Llama-based) - Fast execution for simple tasks
+
+---
+
+## Execution Modes
+
+### Mode 1: Task-Category Parallelization
+Split different task categories across specialized models.
+
+**Best for:** Projects with clearly separated domains (frontend, backend, testing, docs)
+
+```yaml
+Strategy:
+  - Frontend Development → claude-sonnet-4 (visual/UI focus)
+  - Backend API → gpt-4-turbo (code efficiency)
+  - Testing Suite → gemini-2.0-flash (test coverage)
+  - Documentation → perplexity-llama-3.1 (research-backed)
+```
+
+### Mode 2: Phased Parallel Execution
+Execute project phases in parallel across models.
+
+**Best for:** Multi-phase projects with dependencies between phases
+
+```yaml
+Strategy:
+  - Phase 1 (Architecture) → claude-opus-4 (all providers)
+  - Phase 2 (Core Features) → Split across 3 providers
+  - Phase 3 (Testing & Polish) → gpt-4o + gemini-2.0-flash
+  - Phase 4 (Documentation) → perplexity-llama-3.1-sonar
+```
+
+### Mode 3: Redundancy & Verification
+Run critical tasks on multiple models for consensus.
+
+**Best for:** Critical code, security fixes, production deployments
+
+```yaml
+Strategy:
+  - Task 1 (Security Fix) → claude-opus-4 + gpt-4o (parallel)
+  - Task 2 (Core Logic) → claude-sonnet-4 + gemini-2.5-pro
+  - Compare results → Merge best solutions
+```
+
+---
+
+## Quick Start Commands
+
+### 1. System Health Check
+```bash
+# Test all providers
+treequest test
+
+# Check working providers
+treequest status
+
+# List available providers
+treequest providers
+```
+
+### 2. Parallel Deployment - Category Mode
+```bash
+# Deploy 4 parallel agents for different categories
+treequest-parallel \
+  --agents=4 \
+  --mode=category \
+  --config=tmlpd-category.yaml
+```
+
+### 3. Parallel Deployment - Phase Mode
+```bash
+# Deploy phases across multiple models
+treequest-parallel \
+  --agents=6 \
+  --mode=phase \
+  --config=tmlpd-phase.yaml
+```
+
+### 4. Verification Mode
+```bash
+# Run task on 3 models for verification
+treequest-parallel \
+  --agents=3 \
+  --mode=verification \
+  --task="Implement authentication system"
+```
+
+---
+
+## Configuration Files
+
+### tmlpd-category.yaml
+```yaml
+deployment:
+  name: "TMLPD Category Parallelization"
+  agents:
+    - id: "frontend-agent"
+      provider: "anthropic"
+      model: "claude-sonnet-4-20250514"
+      focus: "UI components, styling, animations"
+      tasks: ["task-frontend-*"]
+
+    - id: "backend-agent"
+      provider: "openai"
+      model: "gpt-4-turbo"
+      focus: "API endpoints, database, business logic"
+      tasks: ["task-backend-*"]
+
+    - id: "testing-agent"
+      provider: "google"
+      model: "gemini-2.0-flash-exp"
+      focus: "Unit tests, integration tests, e2e tests"
+      tasks: ["task-testing-*"]
+
+    - id: "docs-agent"
+      provider: "perplexity"
+      model: "llama-3.1-sonar-large-128k-online"
+      focus: "Documentation, README files, API docs"
+      tasks: ["task-docs-*"]
+
+  coordination:
+    sync_interval: 300  # 5 minutes
+    conflict_resolution: "timestamp"
+    merge_strategy: "git-merge"
+
+  output:
+    directory: ".tmlpd-output"
+    logs: "tmlpd-logs.json"
+    metrics: "tmlpd-metrics.json"
+```
+
+### tmlpd-phase.yaml
+```yaml
+deployment:
+  name: "TMLPD Phase Parallelization"
+  agents:
+    - id: "phase1-architect"
+      provider: "anthropic"
+      model: "claude-opus-4"
+      phase: 1
+      focus: "Architecture design, project structure"
+      tasks: ["phase-1-*"]
+
+    - id: "phase2-feature-1"
+      provider: "anthropic"
+      model: "claude-sonnet-4"
+      phase: 2
+      focus: "Core features implementation"
+      tasks: ["phase-2-core-*"]
+
+    - id: "phase2-feature-2"
+      provider: "openai"
+      model: "gpt-4-turbo"
+      phase: 2
+      focus: "Secondary features implementation"
+      tasks: ["phase-2-secondary-*"]
+
+    - id: "phase3-testing"
+      provider: "google"
+      model: "gemini-2.5-pro"
+      phase: 3
+      focus: "Comprehensive testing suite"
+      tasks: ["phase-3-*"]
+
+    - id: "phase4-deployment"
+      provider: "anthropic"
+      model: "claude-sonnet-4"
+      phase: 4
+      focus: "Deployment preparation, CI/CD"
+      tasks: ["phase-4-*"]
+
+    - id: "phase4-documentation"
+      provider: "perplexity"
+      model: "llama-3.1-sonar-large-128k-online"
+      phase: 4
+      focus: "Final documentation, user guides"
+      tasks: ["phase-4-docs-*"]
+
+  coordination:
+    phase_gating: true
+    wait_for_dependencies: true
+    sync_interval: 180
+
+  output:
+    directory: ".tmlpd-phases"
+    logs: "phase-logs.json"
+```
+
+---
+
+## Implementation Workflow
+
+### Step 1: Pre-Deployment Analysis
+```bash
+# Analyze project complexity and task count
+cd /path/to/project
+
+# If using TaskMaster
+task-master analyze-complexity --research
+
+# Get task count
+task-master list --status=pending | wc -l
+```
+
+### Step 2: Create Deployment Plan
+```bash
+# Generate deployment config based on project
+cat > .tmlpd-plan.yaml << EOF
+deployment:
+  project_name: "brain-spark-analysis"
+  total_tasks: 25
+  parallel_strategy: "category"
+  agents_required: 4
+  estimated_duration: "45 minutes"
+EOF
+```
+
+### Step 3: Deploy Parallel Agents
+```bash
+# Launch parallel deployment
+treequest-parallel \
+  --config=.tmlpd-plan.yaml \
+  --background \
+  --monitor
+```
+
+### Step 4: Monitor Execution
+```bash
+# Real-time monitoring
+watch -n 10 'treequest status'
+
+# View agent logs
+tail -f .tmlpd-output/tmlpd-logs.json
+
+# Check progress
+treequest-parallel --status
+```
+
+### Step 5: Merge & Validate
+```bash
+# Merge all agent outputs
+treequest-parallel --merge
+
+# Run validation
+treequest-parallel --validate
+
+# Generate completion report
+treequest-parallel --report > completion-report.md
+```
+
+---
+
+## Advanced Features
+
+### Dynamic Load Balancing
+```yaml
+coordination:
+  load_balancing: "dynamic"
+  rebalance_interval: 120  # 2 minutes
+  overload_threshold: 3     # tasks pending
+  strategy: "work-stealing"
+```
+
+### Cost Optimization
+```yaml
+cost_optimization:
+  enabled: true
+  budget_limit: 50.00  # USD
+  prefer_cached: true
+  fallback_models:
+    - provider: "groq"
+      model: "llama-3.3-70b-versatile"
+      cost_multiplier: 0.1
+```
+
+### Error Recovery
+```yaml
+error_handling:
+  retry_policy: "exponential_backoff"
+  max_retries: 3
+  fallback_on_failure: true
+  checkpoint_interval: 600  # 10 minutes
+```
+
+---
+
+## Integration with Brain Spark Project
+
+### Example: Brain Spark Mobile App Enhancement
+```bash
+cd ~/brain-spark-analysis-project
+
+# Deploy for UI enhancement tasks
+treequest-parallel \
+  --agents=5 \
+  --mode=category \
+  --tasks="task-ui-*,task-accessibility-*,task-i18n-*" \
+  --providers="anthropic,openai,google,groq,cerebras" \
+  --background
+
+# Monitor progress
+treequest-parallel --monitor --refresh=30
+```
+
+### TaskMaster Integration
+```bash
+# Combine with TaskMaster for project management
+task-master parse-prd .taskmaster/docs/ui_enhancement_prd.txt
+task-master expand --all --research
+
+# Deploy TreeQuest agents on expanded tasks
+treequest-parallel \
+  --source=taskmaster \
+  --agents=4 \
+  --config=.tmlpd-brainspark.yaml
+```
+
+---
+
+## Performance Benchmarks
+
+### Expected Performance Gains
+
+| Scenario | Sequential Time | Parallel Time (4 agents) | Speedup |
+|----------|----------------|--------------------------|---------|
+| 25 tasks, mixed complexity | 120 min | 35 min | **3.4x** |
+| 50 tasks, research-heavy | 240 min | 65 min | **3.7x** |
+| 100 tasks, full-stack dev | 480 min | 120 min | **4.0x** |
+
+### Cost Optimization
+
+**Optimal Provider Selection:**
+- Simple tasks (tests, docs) → Groq/Cerebras (10x cheaper)
+- Complex reasoning → Claude Sonnet (balanced cost/performance)
+- Critical architecture → Claude Opus (quality over cost)
+- Research tasks → Perplexity Sonar (web access included)
+
+---
+
+## Troubleshooting
+
+### Issue: Agent Failure
+```bash
+# Check failed agent logs
+treequest-parallel --logs --agent=frontend-agent
+
+# Restart failed agent
+treequest-parallel --restart-agent=frontend-agent
+
+# Resume from checkpoint
+treequest-parallel --resume-checkpoint
+```
+
+### Issue: Resource Exhaustion
+```bash
+# Reduce agent count
+treequest-parallel --config=.tmlpd-plan.yaml --agents=2
+
+# Enable rate limiting
+treequest-parallel --rate-limit=5  # requests per minute
+```
+
+### Issue: Merge Conflicts
+```bash
+# Auto-resolve with timestamp strategy
+treequest-parallel --merge-conflicts --strategy=timestamp
+
+# Manual resolution
+treequest-parallel --merge-conflicts --strategy=manual
+```
+
+---
+
+## Best Practices
+
+1. **Start Small:** Begin with 2-3 agents, scale up based on success
+2. **Monitor Closely:** Use real-time monitoring in first deployments
+3. **Set Budgets:** Always configure cost limits to prevent overruns
+4. **Use Checkpoints:** Enable checkpointing for long-running tasks
+5. **Validate Outputs:** Always run validation after parallel execution
+6. **Document Strategy:** Keep deployment configs in version control
+7. **Profile First:** Run test deployment to measure actual throughput
+
+---
+
+## Example Commands for Brain Spark
+
+### Quick 4-Agent Deployment
+```bash
+cd ~/brain-spark-analysis-project
+
+# Deploy for pending testing tasks
+treequest-parallel \
+  --agents=4 \
+  --mode=category \
+  --filter="status=pending,category=testing" \
+  --providers="anthropic,openai,google,groq" \
+  --output=.tmlpd-output \
+  --background
+
+# Check progress
+treequest-parallel --status
+```
+
+### Research & Documentation Sprint
+```bash
+# Deploy research-focused agents
+treequest-parallel \
+  --agents=3 \
+  --mode=specialized \
+  --focus="research,documentation,analysis" \
+  --providers="perplexity,google,anthropic" \
+  --budget=30.00 \
+  --timeout=3600
+```
+
+---
+
+## Monitoring & Metrics
+
+### Real-Time Dashboard
+```bash
+# Launch monitoring dashboard
+treequest-parallel --dashboard
+
+# View metrics
+treequest-parallel --metrics
+
+# Export performance data
+treequest-parallel --export-metrics > metrics-$(date +%s).json
+```
+
+### Key Metrics Tracked
+- **Tasks Completed:** Total tasks finished per agent
+- **Average Latency:** Response time per provider
+- **Cost Accumulation:** Running cost total
+- **Error Rate:** Failed task percentage
+- **Throughput:** Tasks per minute per agent
+
+---
+
+## Exit Strategy
+
+### Graceful Shutdown
+```bash
+# Stop all agents gracefully
+treequest-parallel --shutdown --graceful
+
+# Force shutdown (emergency)
+treequest-parallel --shutdown --force
+
+# Save partial progress
+treequest-parallel --save-checkpoint
+```
+
+### Resume Operations
+```bash
+# Resume from last checkpoint
+treequest-parallel --resume
+
+# Resume specific agent
+treequest-parallel --resume-agent=frontend-agent
+```
+
+---
+
+## Configuration Templates
+
+Copy these templates to your project for quick deployment:
+
+```bash
+# Copy category template
+cp ~/.claude/skills/tmlpd-category.yaml ~/brain-spark-analysis-project/
+
+# Copy phase template
+cp ~/.claude/skills/tmlpd-phase.yaml ~/brain-spark-analysis-project/
+
+# Copy monitoring config
+cp ~/.claude/skills/tmlpd-monitoring.yaml ~/brain-spark-analysis-project/
+```
+
+---
+
+## Support & Resources
+
+- **TreeQuest Docs:** https://github.com/treequest-ai/treequest
+- **Issue Tracker:** https://github.com/treequest-ai/issues
+- **Config Examples:** `~/.config/treequest/examples/`
+- **Logs:** `~/.treequest/logs/`
+
+---
+
+**Version:** 1.0.0
+**Last Updated:** 2026-01-01
+**Maintained By:** TreeQuest AI Community