Restructure: Move parser to codai.models, add templates, update imports

codai/__init__.py

 # codai module - AI model parsing utilities
-from .model_parser import (
+from .models.parser import (
     ModelParserDispatcher,
     BaseParser,
     QwenParser,
@@ -14,6 +14,8 @@ from .model_parser import (
     ApexBig50Parser,
 )
 
+from .models.templates import AgenticTemplateManager
+
 __all__ = [
     'ModelParserDispatcher',
     'BaseParser',
@@ -27,4 +29,5 @@ __all__ = [
     'GrokParser',
     'PhiParser',
     'ApexBig50Parser',
+    'AgenticTemplateManager',
 ]

codai/models/__init__.py

+# codai.models - Model parsing and templates
+from .parser import (
+    ModelParserDispatcher,
+    BaseParser,
+    QwenParser,
+    DeepSeekParser,
+    LlamaParser,
+    MistralParser,
+    ClaudeParser,
+    CommandRParser,
+    GemmaParser,
+    GrokParser,
+    PhiParser,
+    ApexBig50Parser,
+)
+
+from .templates import AgenticTemplateManager
+
+__all__ = [
+    'ModelParserDispatcher',
+    'BaseParser',
+    'QwenParser',
+    'DeepSeekParser',
+    'LlamaParser',
+    'MistralParser',
+    'ClaudeParser',
+    'CommandRParser',
+    'GemmaParser',
+    'GrokParser',
+    'PhiParser',
+    'ApexBig50Parser',
+    'AgenticTemplateManager',
+]

codai/models/templates.py

+"""
+Agentic Template Manager - Automates prompt injection for agentic behavior.
+Supports the 'Big 10' with specific triggers for tool-calling.
+"""
+
+import re
+
+
+class AgenticTemplateManager:
+    """
+    Automates prompt injection to force models into an Agentic 'Thought-Action' loop.
+    Supports the 'Big 10' with specific triggers for tool-calling.
+    """
+    
+    FAMILIES = {
+        "qwen": {"name": "Qwen", "prefix": "<|im_start|>", "suffix": "<|im_end|>\n", "thought_tag": "<|thought|>", "call_tag": "<tool_call>"},
+        "llama3": {"name": "Llama-3", "prefix": "<|start_header_id|>", "suffix": "<|end_header_id|>\n\n", "thought_tag": "<thought>", "call_tag": "<tool_call>"},
+        "deepseek": {"name": "DeepSeek", "prefix": "<｜", "suffix": "｜>\n", "thought_tag": "<think>", "call_tag": "<tool_call>"},
+        "mistral": {"name": "Mistral", "user": "[INST] ", "bot": " [/INST]", "thought_tag": "Thought: ", "call_tag": "Action: "},
+        "anthropic": {"name": "Claude", "user": "\n\nHuman: ", "bot": "\n\nAssistant: ", "thought_tag": "<thinking>", "call_tag": "<tool_calls>"},
+        "gemma": {"name": "Gemma", "user": "<start_of_turn>user\n", "bot": "<start_of_turn>model\n", "end": "<end_of_turn>\n"},
+        "phi3": {"name": "Phi-3", "prefix": "<|", "suffix": "|>\n", "end": "<|end|>\n"},
+        "cohere": {"name": "Command-R", "user": "<|USER_TOKEN|>", "bot": "<|CHATBOT_TOKEN|>", "thought_tag": "<thought>"},
+        "yi": {"name": "Yi", "prefix": "<|im_start|>", "suffix": "<|im_end|>\n"},
+        "openai": {"name": "GPT"}
+    }
+
+    def __init__(self, model_name: str):
+        self.model_name = model_name.lower()
+        self.family_key = self._detect_family()
+        self.config = self.FAMILIES.get(self.family_key, self.FAMILIES["qwen"])
+
+    def _detect_family(self):
+        mapping = {
+            "qwen": "qwen", "llama-3": "llama3", "deepseek": "deepseek",
+            "mistral": "mistral", "mixtral": "mistral", "claude": "anthropic",
+            "gemma": "gemma", "phi-3": "phi3", "command": "cohere", "yi": "yi"
+        }
+        for k, v in mapping.items():
+            if k in self.model_name: return v
+        return "openai" if "gpt" in self.model_name else "qwen"
+
+    def get_agent_system_prompt(self, base_prompt: str) -> str:
+        """Injects agentic instructions specific to the model's strengths."""
+        agent_addon = (
+            "\n\nCRITICAL: You are an agent with access to tools. "
+            f"Use the {self.config.get('thought_tag', 'Thought:')} tag to reason step-by-step "
+            "before providing a tool call. If you have enough info, provide the final answer."
+        )
+        return f"{base_prompt}{agent_addon}"
+
+    def format_for_inference(self, messages: list) -> str:
+        """Constructs the prompt string and forces the 'Thought' start."""
+        if self.family_key == "openai": return messages
+        
+        prompt = ""
+        f = self.config
+        
+        for m in messages:
+            role, content = m['role'], m['content']
+            
+            # ChatML Style (Qwen, Llama3, Yi, DeepSeek)
+            if "prefix" in f:
+                prompt += f"{f['prefix']}{role}{f['suffix']}{content}"
+                if "end" in f: prompt += f["end"]
+                else: prompt += f.get("suffix", "\n")
+            
+            # Instruction Style (Mistral, Gemma)
+            elif self.family_key == "mistral":
+                if role == "user": prompt += f"{f['user']}{content}{f['bot']}"
+                elif role == "assistant": prompt += f" {content} "
+        
+        # THE AGENTIC PUSH: Force the assistant to start with a THOUGHT
+        thought_trigger = f.get("thought_tag", "Thought: ")
+        if "prefix" in f:
+            prompt += f"{f['prefix']}assistant{f['suffix']}{thought_trigger}"
+        else:
+            prompt += f"{f.get('bot', 'Assistant: ')}{thought_trigger}"
+            
+        return prompt

coderai

@@ -29,7 +29,7 @@ from pydantic_core import PydanticCustomError
 from threading import Thread
 
 # Import codai module for enhanced tool call parsing
-from codai import ModelParserDispatcher
+from codai.models import ModelParserDispatcher
 # Per-model semaphores for request concurrency control
 model_semaphores: dict = {}
 load_mode = {"mode": "ondemand"}  # Track load mode globally