Add support for open router models

DOCUMENTATION.md

@@ -144,7 +144,7 @@ git-smart-squash [OPTIONS]
 | Option | Short | Description | Default |
 |--------|-------|-------------|---------|
 | `--base BASE` | | Base branch to compare against | Config file or `main` |
-| `--ai-provider PROVIDER` | | AI provider to use (`local`, `openai`, `anthropic`, `gemini`) | Config file or `local` |
+| `--ai-provider PROVIDER` | | AI provider to use (`local`, `openai`, `anthropic`, `gemini`, `openrouter`) | Config file or `local` |
 | `--model MODEL` | | Specific model to use | Provider default |
 | `--config PATH` | | Path to configuration file | Auto-detected |
 | `--auto-apply` | | Apply changes without confirmation prompt | Config file or `false` |
@@ -269,6 +269,29 @@ export GEMINI_API_KEY="..."
 
 **Pricing:** ~$0.01 per typical use
 
+### OpenRouter
+
+**Advantages:**
+- Access to 100+ models from multiple providers through a single API
+- Competitive pricing with flexible model selection
+- Unified interface for various AI providers
+- Fast response times with global infrastructure
+- No vendor lock-in - switch models easily
+
+**Setup:**
+```bash
+export OPENROUTER_API_KEY="sk-or-v1-..."
+```
+
+**Models:**
+- `openai/gpt-5` (default) - Latest GPT-5 model with excellent performance
+- `anthropic/claude-3.5-sonnet` - Claude Sonnet through OpenRouter
+- `google/gemini-pro-1.5` - Gemini Pro through OpenRouter
+- `meta-llama/llama-3.1-405b-instruct` - Llama 3.1 through OpenRouter
+- Any other model available on OpenRouter - Specify with `--model` flag
+
+**Pricing:** Variable based on selected model (~$0.005 - $0.03 per typical use)
+
 ## Configuration
 
 ### Configuration Hierarchy
@@ -284,7 +307,7 @@ Git Smart Squash looks for configuration in this order:
 ```yaml
 # AI Provider Settings
 ai:
-  provider: local  # or openai, anthropic, gemini
+  provider: local  # or openai, anthropic, gemini, openrouter
   model: devstral  # optional, uses provider default if not set
                    # You can specify ANY model supported by the provider
   instructions: |  # Optional custom instructions for AI
@@ -316,6 +339,7 @@ auto_apply: false  # Set to true to apply commits without confirmation
 | `OPENAI_API_KEY` | OpenAI API key |
 | `ANTHROPIC_API_KEY` | Anthropic API key |
 | `GEMINI_API_KEY` | Google Gemini API key |
+| `OPENROUTER_API_KEY` | OpenRouter API key |
 | `OLLAMA_HOST` | Ollama server URL (default: http://localhost:11434) |
 
 ## How It Works
@@ -405,6 +429,11 @@ git-smart-squash --ai-provider anthropic --model claude-3-haiku-20240307
 git-smart-squash --ai-provider gemini --model gemini-1.5-flash
 git-smart-squash --ai-provider gemini --model gemini-ultra
 
+# OpenRouter examples
+git-smart-squash --ai-provider openrouter --model openai/gpt-5
+git-smart-squash --ai-provider openrouter --model anthropic/claude-3.5-sonnet
+git-smart-squash --ai-provider openrouter --model google/gemini-pro-1.5
+
 # Local Ollama examples
 git-smart-squash --model codellama:13b
 git-smart-squash --model mistral:latest

README.md

@@ -54,6 +54,12 @@ export ANTHROPIC_API_KEY="your-key"
 export GEMINI_API_KEY="your-key"
 ```
 
+**Option C: OpenRouter (100+ models)**
+```bash
+export OPENROUTER_API_KEY="sk-or-v1-your-key"
+# Then run with: git-smart-squash --ai-provider openrouter
+```
+
 ### 3. Run
 
 ```bash
@@ -68,7 +74,7 @@ That's it. Review the plan and approve.
 | Parameter | Description | Default |
 |-----------|-------------|---------|
 | `--base` | Base branch to compare against | Config file or `main` |
-| `--ai-provider` | AI provider to use (openai, anthropic, local, gemini) | Configured in settings |
+| `--ai-provider` | AI provider to use (openai, anthropic, local, gemini, openrouter) | Configured in settings |
 | `--model` | Specific AI model to use (see recommended models below) | Provider default |
 | `--config` | Path to custom configuration file | `.git-smart-squash.yml` or `~/.git-smart-squash.yml` |
 | `--auto-apply` | Apply commit plan without confirmation prompt | `false` |
@@ -82,6 +88,7 @@ That's it. Review the plan and approve.
 - **OpenAI**: `gpt-4.1` (default)
 - **Anthropic**: `claude-sonnet-4-20250514` (default)
 - **Gemini**: `gemini-2.5-pro` (default)
+- **OpenRouter**: `openai/gpt-5` (default)
 - **Local/Ollama**: `devstral` (default)
 
 ### Model Selection
@@ -91,6 +98,10 @@ git-smart-squash --ai-provider openai --model gpt-4.1-mini
 
 # For local Ollama
 git-smart-squash --ai-provider local --model llama-3.1
+
+# For OpenRouter (access to 100+ models)
+git-smart-squash --ai-provider openrouter --model openai/gpt-5
+git-smart-squash --ai-provider openrouter --model anthropic/claude-3.5-sonnet
 ```
 
 ## Custom Instructions
@@ -130,6 +141,7 @@ git-smart-squash --base develop
 ### "I want to use a specific AI provider"
 ```bash
 git-smart-squash --ai-provider openai
+git-smart-squash --ai-provider openrouter  # Access to 100+ models
 ```
 
 ## Safety
@@ -157,6 +169,7 @@ git reset --hard your-branch-backup-[timestamp]
 | **OpenAI** | ~$0.01 | Cloud |
 | **Anthropic** | ~$0.01 | Cloud |
 | **Gemini** | ~$0.01 | Cloud |
+| **OpenRouter** | ~$0.005-$0.03 | Cloud |
 
 ## Advanced Configuration (Optional)
 

git_smart_squash/ai/providers/simple_unified.py

@@ -17,7 +17,8 @@ class UnifiedAIProvider:
         'local': 32000,       # Ollama hard maximum
         'openai': 1000000,    # OpenAI hard maximum (1M tokens)
         'gemini': 1000000,    # Gemini hard maximum (1M tokens)
-        'anthropic': 200000   # Anthropic hard maximum (200k tokens)
+        'anthropic': 200000,  # Anthropic hard maximum (200k tokens)
+        'openrouter': 1000000 # OpenRouter hard maximum (1M tokens)
     }
 
     # Conservative defaults
@@ -177,6 +178,8 @@ def generate(self, prompt: str) -> str:
             return self._generate_anthropic(prompt)
         elif self.provider_type == "gemini":
             return self._generate_gemini(prompt)
+        elif self.provider_type == "openrouter":
+            return self._generate_openrouter(prompt)
         else:
             raise ValueError(f"Unsupported provider: {self.provider_type}")
 
@@ -439,4 +442,61 @@ def _generate_gemini(self, prompt: str) -> str:
         except ImportError:
             raise Exception("Google Generative AI library not installed. Run: pip install google-generativeai")
         except Exception as e:
-            raise Exception(f"Google Gemini generation failed: {e}")
+            raise Exception(f"Google Gemini generation failed: {e}")
+
+
+    def _generate_openrouter(self, prompt: str) -> str:
+        """Generate using OpenRouter API with structured output enforcement."""
+        try:
+            import openai
+
+            api_key = os.getenv('OPENROUTER_API_KEY')
+            if not api_key:
+                raise Exception("OPENROUTER_API_KEY environment variable not set")
+
+            # Calculate dynamic parameters
+            params = self._calculate_dynamic_params(prompt)
+
+            # Use provider-level context limit
+            model_context_limit = self.PROVIDER_MAX_CONTEXT_TOKENS.get('openrouter', 1000000)
+
+            # Check if prompt exceeds model context limit
+            if params["prompt_tokens"] > model_context_limit - 1000:  # Reserve 1000 for response
+                raise Exception(f"Prompt ({params['prompt_tokens']} tokens) exceeds {self.config.ai.model} context limit ({model_context_limit}). Consider reducing diff size.")
+
+            # Warn if prompt is large but manageable
+            if params["prompt_tokens"] > model_context_limit * 0.7:
+                logger.warning(f"Large prompt ({params['prompt_tokens']} tokens) approaching {self.config.ai.model} context limit.")
+
+            client = openai.OpenAI(
+                base_url="https://openrouter.ai/api/v1",
+                api_key=api_key
+            )
+
+            response = client.chat.completions.create(
+                model=self.config.ai.model,
+                messages=[{"role": "user", "content": prompt}],
+                max_tokens=self.MAX_PREDICT_TOKENS,
+                temperature=0.7,
+                response_format={
+                    "type": "json_schema",
+                    "json_schema": {
+                        "name": "commit_plan",
+                        "schema": self.COMMIT_SCHEMA,
+                        "strict": True
+                    }
+                }
+            )
+
+            # Check if response was truncated
+            if response.choices[0].finish_reason == "length":
+                logger.warning(f"OpenAI response truncated at {self.MAX_PREDICT_TOKENS} tokens. Consider reducing diff size.")
+
+            # Return the full structured response
+            content = response.choices[0].message.content
+            return content  # OpenAI already returns the correct format with json_schema
+
+        except ImportError:
+            raise Exception("OpenAI library not installed. Run: pip install openai")
+        except Exception as e:
+            raise Exception(f"OpenAI generation failed: {e}")

git_smart_squash/cli.py

@@ -79,7 +79,7 @@ def create_parser(self) -> argparse.ArgumentParser:
 
         parser.add_argument(
             '--ai-provider',
-            choices=['openai', 'anthropic', 'local', 'gemini'],
+            choices=['openai', 'anthropic', 'local', 'gemini', 'openrouter'],
             help='AI provider to use'
         )
 

git_smart_squash/simple_config.py

@@ -52,6 +52,7 @@ def _get_default_model(self, provider: str) -> str:
             'openai': 'gpt-4.1',
             'anthropic': 'claude-sonnet-4-20250514',  # Claude Sonnet 4 model
             'gemini': 'gemini-2.5-pro'  # Gemini 2.5 Pro model
+            'openrouter': 'openai/gpt-5'  # OpenAI GPT-5 model
         }
         return defaults.get(provider, 'devstral')