Multi-AI Orchestrator Skill
Description
This skill manages multiple AI services (Codex, Gemini, Claude Code) running on different machines (Linux, Docker, Mac). It handles task routing, token budget management, and service health monitoring.
Capabilities
- •Service Registry: Register and manage multiple AI services
- •Task Routing: Route tasks to appropriate services based on capabilities
- •Token Management: Track and control token usage across all services
- •Health Monitoring: Monitor service availability and performance
- •Result Aggregation: Collect and merge results from multiple services
Available Commands
List Services
bash
python scripts/service_registry.py list [--type <service_type>]
Lists all registered services with their status.
Register Service
bash
python scripts/service_registry.py register --config <config_file>
Register a new service from configuration file.
Route Task
bash
python scripts/router.py --task "<task>" --type <task_type> [--lang <language>] [--domain <domain>]
Route a task to appropriate service(s).
Check Token Budget
bash
python scripts/token_manager.py status [--service <service_id>]
Check token usage and remaining budget.
Health Check
bash
python scripts/service_registry.py health [--id <service_id>]
Check health status of services.
Usage Examples
Route a coding task
code
@orchestrator route --task "Implement user authentication API" --type code_generation --lang python
Check all services
code
@orchestrator list
Get token usage report
code
@orchestrator tokens --report daily
Configuration
Services are configured in config/services/ directory.
Routing rules are in config/routing/task-routing.yaml.
Token budgets are in config/token-budget/per-service-limits.yaml.
Security
- •External services return unified diff patches only
- •No direct filesystem writes from external services
- •Sensitive files (.env, secrets/) are never sent to external services
- •All changes are reviewed and applied by local Claude Code