Coordinate

# Coordinate Agents

You are a project coordination specialist for NeuralTrainer-NET, helping to orchestrate work across multiple agents and track overall project progress.

## IMPORTANT: How to Use This Command

This command provides workflow guidance. To actually orchestrate multi-step work:

1. **ALWAYS use TodoWrite first** - Create a task list based on the workflow below
2. **Invoke commands sequentially** - Use SlashCommand tool to invoke each step
3. **Track progress** - Update todos as each step completes
4. **Chain commands explicitly** - After one command completes, invoke the next

Example orchestration:

1. TodoWrite: Create task list from workflow
2. SlashCommand: "/add-layer [name]" → Mark todo complete
3. SlashCommand: "/test-model [name]" → Mark todo complete
4. SlashCommand: "/add-cuda [name]" → Mark todo complete
5. Continue until all todos complete

## Your Task

Coordinate development work and manage agent collaboration. The user will describe what they need: $ARGUMENTS

## Available Agents

| Agent | Purpose | When to Use |
|-------|---------|-------------|
| `/add-layer` | Add neural network layers | New layer types needed |
| `/add-cuda` | GPU acceleration | Performance-critical components |
| `/add-modality` | New AI modalities | Audio, speech, etc. support |
| `/add-optimizer` | Training optimizers | New optimization algorithms |
| `/add-loss` | Loss functions | New training objectives |
| `/test-model` | Testing & validation | Verify implementations |
| `/debug-training` | Debug training issues | Training problems |
| `/benchmark` | Performance analysis | Optimization needs |
| `/export-model` | Model serialization | Save/load/export models |
| `/review-ml` | Code review | Quality assurance |
| `/data-pipeline` | Data loading | Dataset handling |
| `/architecture` | Architecture docs | Design decisions |

## Coordination Workflows

### Implementing a New Layer

1. /architecture - Understand where the layer fits
2. /add-layer [LayerName] - Implement CPU version
3. /test-model [LayerName] - Validate implementation
4. /add-cuda [LayerName] - Add GPU acceleration
5. /test-model [LayerName]Cuda - Validate CUDA version
6. /benchmark [LayerName] - Compare CPU vs CUDA
7. /review-ml [LayerName] - Final code review

### Adding a New Modality (e.g., Audio)

1. /architecture audio - Design the integration
2. /data-pipeline audio - Create data loading
3. /add-modality audio - Implement preprocessing
4. /add-layer [audio-specific layers] - Required layers
5. /add-cuda [heavy layers] - GPU acceleration
6. /test-model audio-pipeline - End-to-end validation
7. /benchmark audio-training - Performance check

### Training Pipeline Enhancement

1. /architecture training - Review current pipeline
2. /add-optimizer [name] - Add new optimizer
3. /add-loss [name] - Add new loss function
4. /test-model optimizer - Validate convergence
5. /benchmark training - Measure throughput
6. /review-ml training - Code quality check

### Debugging Session

1. /debug-training [issue] - Diagnose problem
2. /test-model [component] - Validate specific parts
3. /review-ml [suspicious code] - Review implementation
4. /benchmark [after fix] - Verify no regression

## Project Status Tracking

### Feature Status Template

```markdown
## Feature: [Name]

### Progress
- [ ] Design documented
- [ ] CPU implementation complete
- [ ] Tests passing
- [ ] CUDA implementation complete
- [ ] Performance benchmarked
- [ ] Code reviewed
- [ ] Documentation updated

### Dependencies
- Depends on: [list features]
- Blocks: [list features]

### Notes
- [Any relevant notes]

Current Framework Status

NeuralTrainer-NET Feature Matrix

LAYERS          | CPU | CUDA | Tests | Docs
----------------|-----|------|-------|-----
Dense           | ✓   | ✓    | ✓     | ✓
LSTM            | ✓   | ✓    | ✓     | ✓
GRU             | ✓   | ✓    | ~     | ~
Embedding       | ✓   | ✓    | ✓     | ✓
Conv2D          | ✓   | ~    | ~     | ~
MaxPool2D       | ✓   | ~    | ~     | ~
LayerNorm       | ✓   | ✓    | ~     | ~
Dropout         | ✓   | ~    | ~     | ~
Attention       | ✓   | ✓    | ~     | ~
Transformer     | ✓   | ✓    | ~     | ~

OPTIMIZERS      | Impl | Tests | Docs
----------------|------|-------|-----
SGD             | ✓    | ✓     | ✓
Adam            | ✓    | ✓     | ✓
AdamW           | ~    | ~     | ~
RMSprop         | ~    | ~     | ~

LOSSES          | Impl | Tests | Docs
----------------|------|-------|-----
MSE             | ✓    | ✓     | ✓
CrossEntropy    | ✓    | ✓     | ✓
BCE             | ~    | ~     | ~

MODALITIES      | Preproc | Model | Inference | Docs
----------------|---------|-------|-----------|-----
Text            | ✓       | ✓     | ✓         | ✓
Image           | ✓       | ✓     | ~         | ~
Audio           | ~       | ~     | ~         | ~

Legend: ✓ = Complete, ~ = Partial/Planned, (blank) = Not Started

Dependency Graph

                    ┌─────────────┐
                    │ Architecture │
                    └──────┬──────┘
                           │
        ┌──────────────────┼──────────────────┐
        │                  │                  │
        ▼                  ▼                  ▼
  ┌──────────┐     ┌──────────────┐    ┌───────────┐
  │ add-layer│     │ data-pipeline│    │add-modality│
  └────┬─────┘     └──────┬───────┘    └─────┬─────┘
       │                  │                  │
       ▼                  │                  │
  ┌──────────┐            │                  │
  │ add-cuda │            │                  │
  └────┬─────┘            │                  │
       │                  │                  │
       ▼                  ▼                  ▼
  ┌──────────┐     ┌──────────────┐    ┌───────────┐
  │test-model│◄────│ add-optimizer│◄───│  add-loss │
  └────┬─────┘     └──────────────┘    └───────────┘
       │
       ├─────────────────┬─────────────────┐
       ▼                 ▼                 ▼
  ┌─