Multimodal Agent Capability Mapper

You are a multimodal AI agent architect. I need you to design a comprehensive capability mapping system for an agent that handles multiple input/output modalities.

## Agent Specifications
**Primary Function:** [describe main agent purpose]
**Target Modalities:** [text/audio/video/image/realtime]
**Deployment Environment:** [production/research/prototype]
**Performance Requirements:** [latency/throughput/accuracy targets]

## Current Implementation
[paste current agent architecture or planned components]

## Create This Capability Map

### 1. Modality Matrix
Create a grid showing:
- Input modalities (rows) × Output modalities (columns)
- Supported combinations with complexity ratings
- Resource requirements for each combination
- Latency expectations per modality pair

### 2. Capability Boundaries
Define clear limits for:
- Maximum input size per modality
- Concurrent modality handling capacity
- Cross-modal reasoning depth
- Context preservation across modality switches

### 3. Fallback Strategies
Design degradation paths for:
- Modality unavailability scenarios
- Resource constraint situations
- Quality degradation handling
- User experience preservation methods

### 4. Integration Architecture
Specify:
- Model orchestration patterns
- Data flow between modality processors
- Shared context management
- State synchronisation mechanisms

### 5. Validation Framework
Create tests for:
- Cross-modal consistency verification
- Performance benchmarks per modality
- End-to-end workflow validation
- Edge case handling verification

### 6. Monitoring Strategy
Define metrics for:
- Per-modality performance tracking
- Cross-modal coherence measurement
- Resource utilisation monitoring
- User experience quality indicators

Include specific test scenarios, performance thresholds, and architectural decision rationale. Focus on practical implementation guidance.