AI Agent Framework
AltSportsLeagues.ai employs a sophisticated multi-agent architecture with 50+ specialized agents organized into composable, trait-based patterns for maximum flexibility and reusability.
Agent Categories
Agent Definition Structure
Every agent follows a standardized YAML frontmatter + Markdown structure with the following properties:
| Property | Type | Description |
|---|---|---|
name | string | Unique identifier (kebab-case) |
description | string | 1-2 sentence summary |
auto_activate_phrases | array | Keywords that trigger this agent |
tools | array | Available tool capabilities |
model | enum | Claude model to use |
category | string | Classification for routing |
tags | array | Searchable metadata |
Quality Standards
All agents adhere to strict quality standards:
| Dimension | Weight |
|---|---|
| Code Quality | 25% |
| Test Coverage | 25% |
| Documentation | 20% |
| Performance | 15% |
| Security | 15% |
Quality Metrics
| Metric | Standard | Measurement |
|---|---|---|
| Code Quality | Production-ready | Linting score greater than 90% |
| Test Coverage | Greater than 80% | Jest/pytest coverage |
| Documentation | Complete | All public APIs documented |
| Type Safety | Greater than 90% | TypeScript/Python typing |
| Performance | Benchmarked | Response time under 500ms |
Orchestration Patterns
Complex tasks leverage multiple agents working together:
- Sequential Pipeline - Agents process in order
- Parallel Fan-out - Multiple agents work simultaneously
- Supervisor Pattern - Orchestrator monitors and coordinates
- Consensus Pattern - Multiple agents vote on output
Creating Custom Agents
See Creating Custom Agents for a complete guide on building your own specialized agents with composable traits.
Related Documentation
- System Specifications - Kiro-style requirements
- Schema Registry - Data model documentation
- Internal Developer Guide - Development workflows