1  #!/usr/bin/env python3
  2  """
  3  Model Router CLI - AI Agent Agnostic interface.
  4  
  5  Treats all AI models the same:
  6      python3 scripts/model_router_cli.py "your question"
  7      python3 scripts/model_router_cli.py --model gemini-flash "quick question"
  8      python3 scripts/model_router_cli.py --status
  9      python3 scripts/model_router_cli.py --consensus "important decision"
 10  
 11  The router automatically:
 12  - Routes to best available model
 13  - Falls back if one fails
 14  - Injects Sovereign OS context
 15  """
 16  
 17  import sys
 18  import os
 19  import json
 20  from pathlib import Path
 21  
 22  # Add core to path
 23  SOVEREIGN_ROOT = Path(__file__).parent.parent
 24  sys.path.insert(0, str(SOVEREIGN_ROOT))
 25  
 26  from core.mesh import ModelRouter, TaskType
 27  
 28  
 29  def print_status(router: ModelRouter):
 30      """Print router status."""
 31      status = router.status()
 32  
 33      print("\n" + "=" * 60)
 34      print("MODEL ROUTER STATUS")
 35      print("=" * 60)
 36  
 37      print(f"\nAvailable models: {', '.join(status['available']) or 'NONE'}")
 38      print(f"Mesh connected: {status['mesh_connected']}")
 39  
 40      print("\nAdapter Details:")
 41      print("-" * 60)
 42  
 43      for name, info in status['adapters'].items():
 44          available = "✓" if info['available'] else "✗"
 45          caps = info['capabilities']
 46          print(f"  {available} {name}")
 47          print(f"      coding={caps['coding']:.1f} research={caps['research']:.1f} "
 48                f"speed={caps['speed']:.1f} cost={caps['cost_efficiency']:.1f}")
 49          print(f"      context={caps['context_window']:,} tokens")
 50  
 51      print("\n" + "=" * 60)
 52  
 53  
 54  def main():
 55      args = sys.argv[1:]
 56  
 57      if not args:
 58          print(__doc__)
 59          print("\nQuick status check:")
 60          router = ModelRouter()
 61          available = router.available_models()
 62          if available:
 63              print(f"  Available: {', '.join(available)}")
 64          else:
 65              print("  No models available. Set GEMINI_API_KEY or ANTHROPIC_API_KEY")
 66          return
 67  
 68      router = ModelRouter()
 69  
 70      # Handle flags
 71      if args[0] == "--status":
 72          print_status(router)
 73          return
 74  
 75      if args[0] == "--consensus":
 76          if len(args) < 2:
 77              print("Usage: --consensus <prompt>")
 78              return
 79          prompt = " ".join(args[1:])
 80          print("\nQuerying multiple models for consensus...\n")
 81          result = router.consensus(prompt)
 82          if "error" in result:
 83              print(f"Error: {result['error']}")
 84              return
 85          print(f"Models queried: {', '.join(result['models_queried'])}")
 86          print(f"Synthesis: {result['synthesis']}")
 87          print("\nIndividual responses:")
 88          for model, resp in result['responses'].items():
 89              print(f"\n--- {model} ---")
 90              print(resp['content'][:500] + "..." if len(resp['content']) > 500 else resp['content'])
 91          return
 92  
 93      # Parse model flag
 94      model = None
 95      task_type = TaskType.CODING
 96      prompt_start = 0
 97  
 98      if args[0] == "--model":
 99          if len(args) < 3:
100              print("Usage: --model <model-name> <prompt>")
101              return
102          model = args[1]
103          prompt_start = 2
104  
105      if args[prompt_start] == "--research":
106          task_type = TaskType.RESEARCH
107          prompt_start += 1
108      elif args[prompt_start] == "--quick":
109          task_type = TaskType.QUICK
110          prompt_start += 1
111  
112      prompt = " ".join(args[prompt_start:])
113  
114      if not prompt:
115          print("Please provide a prompt.")
116          return
117  
118      # Query
119      print(f"\nRouting to {'best available' if not model else model}...\n")
120      response = router.query(
121          prompt,
122          task_type=task_type,
123          model=model,
124      )
125  
126      print(f"[{response.model_name}] ({response.latency_ms}ms)")
127      print("-" * 60)
128      print(response.content)
129  
130  
131  if __name__ == "__main__":
132      main()