hanzo 0.3.28__tar.gz → 0.3.29__tar.gz

{hanzo-0.3.28 → hanzo-0.3.29}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hanzo
-Version: 0.3.28
+Version: 0.3.29
 Summary: Hanzo AI - Complete AI Infrastructure Platform with CLI, Router, MCP, and Agent Runtime
 Project-URL: Homepage, https://hanzo.ai
 Project-URL: Repository, https://github.com/hanzoai/python-sdk

{hanzo-0.3.28 → hanzo-0.3.29}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "hanzo"
-version = "0.3.28"
+version = "0.3.29"
 description = "Hanzo AI - Complete AI Infrastructure Platform with CLI, Router, MCP, and Agent Runtime"
 authors = [
     {name = "Hanzo AI", email = "dev@hanzo.ai"},

{hanzo-0.3.28 → hanzo-0.3.29}/src/hanzo/__init__.py

@@ -1,6 +1,6 @@
 """Hanzo - Complete AI Infrastructure Platform with CLI, Router, MCP, and Agent Runtime."""
 
-__version__ = "0.3.28"
+__version__ = "0.3.29"
 __all__ = ["main", "cli", "__version__"]
 
 from .cli import cli, main

{hanzo-0.3.28 → hanzo-0.3.29}/src/hanzo/interactive/enhanced_repl.py

@@ -645,8 +645,25 @@ class EnhancedHanzoREPL:
             if success:
                 self.console.print(output)
             else:
-                # Fallback to regular model
-                self.console.print(f"[yellow]{self.current_tool.display_name} failed, trying cloud model...[/yellow]")
+                # Show error and try fallback
+                self.console.print(f"[red]Error: {output}[/red]")
+                
+                # Try to find next available tool
+                if self.tool_detector and self.tool_detector.detected_tools:
+                    for fallback_tool in self.tool_detector.detected_tools:
+                        if fallback_tool != self.current_tool:
+                            self.console.print(f"[yellow]Trying {fallback_tool.display_name}...[/yellow]")
+                            success, output = self.tool_detector.execute_with_tool(fallback_tool, message)
+                            if success:
+                                self.console.print(output)
+                                # Suggest switching to working tool
+                                self.console.print(f"\n[dim]Tip: Switch to {fallback_tool.display_name} with /model {fallback_tool.name}[/dim]")
+                                return
+                            else:
+                                self.console.print(f"[red]{fallback_tool.display_name} also failed[/red]")
+                
+                # Final fallback to cloud model
+                self.console.print(f"[yellow]Falling back to cloud model...[/yellow]")
                 await self.execute_command("ask", f"--cloud --model gpt-3.5-turbo {message}")
         else:
             # Use regular model through hanzo ask

{hanzo-0.3.28 → hanzo-0.3.29}/src/hanzo/tools/detector.py

@@ -170,22 +170,46 @@ class ToolDetector:
             try:
                 response = httpx.get(tool.api_endpoint, timeout=1.0)
                 if response.status_code == 200:
-                    tool.detected = True
-                    tool.version = "Running"
-                    
-                    # Special handling for Hanzo services
+                    # For Hanzo Node, verify it can actually handle chat completions
                     if tool.name == "hanzod":
-                        # Check if models are loaded
                         try:
-                            models_response = httpx.get("http://localhost:8000/models", timeout=1.0)
-                            if models_response.status_code == 200:
-                                models = models_response.json()
-                                if models:
-                                    tool.version = f"Running ({len(models)} models)"
+                            # Check if the chat completions endpoint works
+                            test_response = httpx.post(
+                                "http://localhost:8000/v1/chat/completions",
+                                json={
+                                    "messages": [{"role": "user", "content": "test"}],
+                                    "model": "test",
+                                    "max_tokens": 1
+                                },
+                                timeout=2.0
+                            )
+                            # Only mark as detected if we get a valid response or specific error
+                            # 404 means the endpoint doesn't exist
+                            if test_response.status_code == 404:
+                                return False
+                            
+                            tool.detected = True
+                            tool.version = "Running (Local AI)"
+                            
+                            # Try to get model info
+                            try:
+                                models_response = httpx.get("http://localhost:8000/v1/models", timeout=1.0)
+                                if models_response.status_code == 200:
+                                    models = models_response.json().get("data", [])
+                                    if models:
+                                        tool.version = f"Running ({len(models)} models)"
+                            except:
+                                pass
+                            
+                            return True
                         except:
-                            pass
-                    
-                    return True
+                            # If chat endpoint doesn't work, node isn't useful
+                            return False
+                    else:
+                        # For other services, just check health endpoint
+                        tool.detected = True
+                        tool.version = "Running"
+                        return True
             except:
                 pass
         
@@ -318,12 +342,13 @@ class ToolDetector:
         try:
             # Special handling for Hanzo services
             if tool.name == "hanzod":
-                # Use the local API directly
+                # Use the local API directly with correct endpoint
                 try:
                     response = httpx.post(
-                        "http://localhost:8000/chat/completions",
+                        "http://localhost:8000/v1/chat/completions",
                         json={
                             "messages": [{"role": "user", "content": prompt}],
+                            "model": "default",  # Use default model
                             "stream": False
                         },
                         timeout=30.0
@@ -331,6 +356,8 @@ class ToolDetector:
                     if response.status_code == 200:
                         result = response.json()
                         return True, result.get("choices", [{}])[0].get("message", {}).get("content", "")
+                    else:
+                        return False, f"Hanzo Node returned {response.status_code}: {response.text}"
                 except Exception as e:
                     return False, f"Hanzo Node error: {e}"