fix: simplify tool instructions to ~300 tokens (was 40k)

Reduce tool instructions from 40k tokens to ~300 tokens:
- List only 3 main tools (read, write, bash) with brief descriptions
- Single concise JSON format example
- Remove verbose formatting and multiple examples
- Only add instructions on first request (no assistant response yet)

This makes tool usage feasible for 8K-32K context models,
especially important for home setups with limited VRAM.

src/api/routes.py

@@ -59,8 +59,34 @@ def format_messages_with_tools(messages: list, tools: Optional[list] = None) ->
     """
     formatted = []
     
-    # Tools are accepted but model responds normally
-    # Server-side tool execution parses the response and executes tools
+    # Check if there are already tool results in the conversation
+    has_tool_results = any(msg.role == "tool" for msg in messages)
+    has_assistant_response = any(msg.role == "assistant" for msg in messages)
+    
+    # Add brief tool instructions if tools are present and no assistant has responded yet
+    if tools and not has_tool_results and not has_assistant_response:
+        tool_instructions = """You have access to these tools:
+
+read: Read a file (filePath)
+write: Write to a file (filePath, content)
+bash: Run a shell command (command)
+
+When you need to use a tool, respond with ONLY this JSON format:
+{"tool_calls": [{"id": "call_1", "type": "function", "function": {"name": "TOOL_NAME", "arguments": "{\\"param\\": \\"value\\"}"}}]}
+
+Do not explain - just output the JSON when using tools."""
+        
+        # Add to system message or create one
+        has_system = False
+        for msg in messages:
+            if msg.role == "system":
+                msg.content = tool_instructions + "\n\n" + (msg.content or "")
+                has_system = True
+                break
+        
+        if not has_system:
+            from api.models import ChatMessage
+            messages.insert(0, ChatMessage(role="system", content=tool_instructions))
     
     for msg in messages:
         role = msg.role