tunacode-cli 0.0.13__tar.gz → 0.0.14__tar.gz

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tunacode-cli
-Version: 0.0.13
+Version: 0.0.14
 Summary: Your agentic CLI developer.
 Author-email: larock22 <noreply@github.com>
 License-Expression: MIT
@@ -58,13 +58,14 @@ Dynamic: license-file
 
 ---
 
-### Recent Updates
+### Recent Updates (v0.0.14)
 
-- **Simplified Setup**: Direct CLI configuration with `--model` and `--key` flags
-- **Enhanced Safety**: Removed `/undo` command in favor of git-based workflows
-- **Cleaner Codebase**: Removed `/init` command and automatic TUNACODE.md generation
-- **Better Onboarding**: No model validation - trust users to provide correct model names
-- **Unified Model Format**: All models use `provider:model-name` format
+- **🔧 JSON Tool Parsing Fallback**: Automatic recovery when API providers fail with structured tool calling
+- **⚡ Enhanced Reliability**: Fixed parameter naming issues that caused tool schema errors
+- **🔄 Configuration Management**: New `/refresh` command to reload config without restart
+- **🧠 Improved ReAct Reasoning**: Enhanced iteration limits (now defaults to 20) and better thought processing
+- **🛠️ New Debug Commands**: `/parsetools` for manual JSON parsing, `/iterations` for controlling reasoning depth
+- **📊 Better Error Recovery**: Multiple fallback mechanisms for various failure scenarios
 
 ### Core Features
 
@@ -75,17 +76,18 @@ Dynamic: license-file
 ### **Multi-Provider Support**
 
 - Anthropic Claude
-- OpenAI GPT
+- OpenAI GPT  
 - Google Gemini
 - OpenRouter (100+ models)
 - Any OpenAI-compatible API
 
 ### **Developer Tools**
 
-- 4 core tools: read_file, write_file, update_file, run_command
+- 6 core tools: bash, grep, read_file, write_file, update_file, run_command
 - MCP (Model Context Protocol) support
 - File operation confirmations with diffs
 - Per-project context guides (TUNACODE.md)
+- JSON tool parsing fallback for API compatibility
 
 </td>
 <td width="50%">
@@ -104,6 +106,7 @@ Dynamic: license-file
 - Async throughout
 - Modular command system
 - Rich UI with syntax highlighting
+- ReAct reasoning patterns
 
 </td>
 </tr>
@@ -268,6 +271,8 @@ Learn more at [modelcontextprotocol.io](https://modelcontextprotocol.io/)
 
 ## Commands Reference
 
+### Core Commands
+
 | Command                          | Description                      |
 | -------------------------------- | -------------------------------- |
 | `/help`                          | Show available commands          |
@@ -281,13 +286,43 @@ Learn more at [modelcontextprotocol.io](https://modelcontextprotocol.io/)
 | `/dump`                          | Show message history (debug)     |
 | `exit`                           | Exit application                 |
 
+### Debug & Recovery Commands
+
+| Command                          | Description                      |
+| -------------------------------- | -------------------------------- |
+| `/thoughts`                      | Toggle ReAct thought display     |
+| `/iterations <1-50>`             | Set max reasoning iterations     |
+| `/parsetools`                    | Parse JSON tool calls manually   |
+| `/fix`                           | Fix orphaned tool calls         |
+| `/refresh`                       | Reload configuration from defaults |
+
+---
+
+## Reliability Features
+
+### JSON Tool Parsing Fallback
+
+TunaCode automatically handles API provider failures with robust JSON parsing:
+
+- **Automatic Recovery**: When structured tool calling fails, TunaCode parses JSON from text responses
+- **Multiple Formats**: Supports inline JSON, code blocks, and complex nested structures
+- **Manual Recovery**: Use `/parsetools` when automatic parsing needs assistance
+- **Visual Feedback**: See `🔧 Recovered using JSON tool parsing` messages during fallback
+
+### Enhanced Error Handling
+
+- **Tool Schema Fixes**: Consistent parameter naming across all tools
+- **Orphaned Tool Call Recovery**: Automatic cleanup with `/fix` command
+- **Configuration Refresh**: Update settings without restart using `/refresh`
+- **ReAct Reasoning**: Configurable iteration limits for complex problem solving
+
 ---
 
 ## Customization
 
 ### Project Guides
 
-Create a `TUNACODE.md` file your project root to customize TunaCode's behavior:
+Create a `TUNACODE.md` file in your project root to customize TunaCode's behavior:
 
 ```markdown
 # Project Guide
@@ -337,13 +372,14 @@ src/tunacode/
 │   └── tool_handler.py   # Tool execution and validation
 │
 ├── services/             # External Services
-│   ├── mcp.py           # Model Context Protocol integration
-│   └── undo_service.py  # Undo operations (beta)
+│   └── mcp.py           # Model Context Protocol integration
 │
 ├── tools/               # AI Agent Tools
 │   ├── base.py         # Tool base classes
+│   ├── bash.py         # Enhanced shell command execution
+│   ├── grep.py         # Parallel content search tool
 │   ├── read_file.py    # File reading tool
-│   ├── run_command.py  # Command execution tool
+│   ├── run_command.py  # Basic command execution tool
 │   ├── update_file.py  # File modification tool
 │   └── write_file.py   # File creation tool
 │
@@ -360,7 +396,7 @@ src/tunacode/
 │   └── validators.py   # Input validation
 │
 ├── utils/              # Utility Functions
-│   ├── bm25.py        # BM25 search algorithm(beta)
+│   ├── bm25.py        # BM25 search algorithm (beta)
 │   ├── diff_utils.py  # Diff generation and formatting
 │   ├── file_utils.py  # File system operations
 │   ├── ripgrep.py     # Code search utilities

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/README.md

@@ -22,13 +22,14 @@
 
 ---
 
-### Recent Updates
+### Recent Updates (v0.0.14)
 
-- **Simplified Setup**: Direct CLI configuration with `--model` and `--key` flags
-- **Enhanced Safety**: Removed `/undo` command in favor of git-based workflows
-- **Cleaner Codebase**: Removed `/init` command and automatic TUNACODE.md generation
-- **Better Onboarding**: No model validation - trust users to provide correct model names
-- **Unified Model Format**: All models use `provider:model-name` format
+- **🔧 JSON Tool Parsing Fallback**: Automatic recovery when API providers fail with structured tool calling
+- **⚡ Enhanced Reliability**: Fixed parameter naming issues that caused tool schema errors
+- **🔄 Configuration Management**: New `/refresh` command to reload config without restart
+- **🧠 Improved ReAct Reasoning**: Enhanced iteration limits (now defaults to 20) and better thought processing
+- **🛠️ New Debug Commands**: `/parsetools` for manual JSON parsing, `/iterations` for controlling reasoning depth
+- **📊 Better Error Recovery**: Multiple fallback mechanisms for various failure scenarios
 
 ### Core Features
 
@@ -39,17 +40,18 @@
 ### **Multi-Provider Support**
 
 - Anthropic Claude
-- OpenAI GPT
+- OpenAI GPT  
 - Google Gemini
 - OpenRouter (100+ models)
 - Any OpenAI-compatible API
 
 ### **Developer Tools**
 
-- 4 core tools: read_file, write_file, update_file, run_command
+- 6 core tools: bash, grep, read_file, write_file, update_file, run_command
 - MCP (Model Context Protocol) support
 - File operation confirmations with diffs
 - Per-project context guides (TUNACODE.md)
+- JSON tool parsing fallback for API compatibility
 
 </td>
 <td width="50%">
@@ -68,6 +70,7 @@
 - Async throughout
 - Modular command system
 - Rich UI with syntax highlighting
+- ReAct reasoning patterns
 
 </td>
 </tr>
@@ -232,6 +235,8 @@ Learn more at [modelcontextprotocol.io](https://modelcontextprotocol.io/)
 
 ## Commands Reference
 
+### Core Commands
+
 | Command                          | Description                      |
 | -------------------------------- | -------------------------------- |
 | `/help`                          | Show available commands          |
@@ -245,13 +250,43 @@ Learn more at [modelcontextprotocol.io](https://modelcontextprotocol.io/)
 | `/dump`                          | Show message history (debug)     |
 | `exit`                           | Exit application                 |
 
+### Debug & Recovery Commands
+
+| Command                          | Description                      |
+| -------------------------------- | -------------------------------- |
+| `/thoughts`                      | Toggle ReAct thought display     |
+| `/iterations <1-50>`             | Set max reasoning iterations     |
+| `/parsetools`                    | Parse JSON tool calls manually   |
+| `/fix`                           | Fix orphaned tool calls         |
+| `/refresh`                       | Reload configuration from defaults |
+
+---
+
+## Reliability Features
+
+### JSON Tool Parsing Fallback
+
+TunaCode automatically handles API provider failures with robust JSON parsing:
+
+- **Automatic Recovery**: When structured tool calling fails, TunaCode parses JSON from text responses
+- **Multiple Formats**: Supports inline JSON, code blocks, and complex nested structures
+- **Manual Recovery**: Use `/parsetools` when automatic parsing needs assistance
+- **Visual Feedback**: See `🔧 Recovered using JSON tool parsing` messages during fallback
+
+### Enhanced Error Handling
+
+- **Tool Schema Fixes**: Consistent parameter naming across all tools
+- **Orphaned Tool Call Recovery**: Automatic cleanup with `/fix` command
+- **Configuration Refresh**: Update settings without restart using `/refresh`
+- **ReAct Reasoning**: Configurable iteration limits for complex problem solving
+
 ---
 
 ## Customization
 
 ### Project Guides
 
-Create a `TUNACODE.md` file your project root to customize TunaCode's behavior:
+Create a `TUNACODE.md` file in your project root to customize TunaCode's behavior:
 
 ```markdown
 # Project Guide
@@ -301,13 +336,14 @@ src/tunacode/
 │   └── tool_handler.py   # Tool execution and validation
 │
 ├── services/             # External Services
-│   ├── mcp.py           # Model Context Protocol integration
-│   └── undo_service.py  # Undo operations (beta)
+│   └── mcp.py           # Model Context Protocol integration
 │
 ├── tools/               # AI Agent Tools
 │   ├── base.py         # Tool base classes
+│   ├── bash.py         # Enhanced shell command execution
+│   ├── grep.py         # Parallel content search tool
 │   ├── read_file.py    # File reading tool
-│   ├── run_command.py  # Command execution tool
+│   ├── run_command.py  # Basic command execution tool
 │   ├── update_file.py  # File modification tool
 │   └── write_file.py   # File creation tool
 │
@@ -324,7 +360,7 @@ src/tunacode/
 │   └── validators.py   # Input validation
 │
 ├── utils/              # Utility Functions
-│   ├── bm25.py        # BM25 search algorithm(beta)
+│   ├── bm25.py        # BM25 search algorithm (beta)
 │   ├── diff_utils.py  # Diff generation and formatting
 │   ├── file_utils.py  # File system operations
 │   ├── ripgrep.py     # Code search utilities
@@ -402,4 +438,4 @@ MIT License - see [LICENSE](LICENSE) file for details.
 
 ## Acknowledgments
 
-TunaCode is a fork of [sidekick-cli](https://github.com/geekforbrains/sidekick-cli). Special thanks to the sidekick-cli team for creating the foundation that made TunaCode possible.
+TunaCode is a fork of [sidekick-cli](https://github.com/geekforbrains/sidekick-cli). Special thanks to the sidekick-cli team for creating the foundation that made TunaCode possible.

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "tunacode-cli"
-version = "0.0.13"
+version = "0.0.14"
 description = "Your agentic CLI developer."
 keywords = ["cli", "agent", "development", "automation"]
 readme = "README.md"

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/src/tunacode/cli/commands.py

@@ -138,6 +138,73 @@ class DumpCommand(SimpleCommand):
         await ui.dump_messages(context.state_manager.session.messages)
 
 
+class ThoughtsCommand(SimpleCommand):
+    """Toggle display of agent thoughts."""
+
+    def __init__(self):
+        super().__init__(
+            CommandSpec(
+                name="thoughts",
+                aliases=["/thoughts"],
+                description="Show or hide agent thought messages",
+                category=CommandCategory.DEBUG,
+            )
+        )
+
+    async def execute(self, args: List[str], context: CommandContext) -> None:
+        state = context.state_manager.session
+        if args:
+            arg = args[0].lower()
+            if arg in {"on", "1", "true"}:
+                state.show_thoughts = True
+            elif arg in {"off", "0", "false"}:
+                state.show_thoughts = False
+            else:
+                await ui.error("Usage: /thoughts [on|off]")
+                return
+        else:
+            state.show_thoughts = not state.show_thoughts
+        status = "ON" if state.show_thoughts else "OFF"
+        await ui.success(f"Thought display {status}")
+
+
+class IterationsCommand(SimpleCommand):
+    """Configure maximum agent iterations for ReAct reasoning."""
+
+    def __init__(self):
+        super().__init__(
+            CommandSpec(
+                name="iterations",
+                aliases=["/iterations"],
+                description="Set maximum agent iterations for complex reasoning",
+                category=CommandCategory.DEBUG,
+            )
+        )
+
+    async def execute(self, args: List[str], context: CommandContext) -> None:
+        state = context.state_manager.session
+        if args:
+            try:
+                new_limit = int(args[0])
+                if new_limit < 1 or new_limit > 50:
+                    await ui.error("Iterations must be between 1 and 50")
+                    return
+                
+                # Update the user config
+                if "settings" not in state.user_config:
+                    state.user_config["settings"] = {}
+                state.user_config["settings"]["max_iterations"] = new_limit
+                
+                await ui.success(f"Maximum iterations set to {new_limit}")
+                await ui.muted("Higher values allow more complex reasoning but may be slower")
+            except ValueError:
+                await ui.error("Please provide a valid number")
+        else:
+            current = state.user_config.get("settings", {}).get("max_iterations", 15)
+            await ui.info(f"Current maximum iterations: {current}")
+            await ui.muted("Usage: /iterations <number> (1-50)")
+
+
 class ClearCommand(SimpleCommand):
     """Clear screen and message history."""
 
@@ -152,8 +219,127 @@ class ClearCommand(SimpleCommand):
         )
 
     async def execute(self, args: List[str], context: CommandContext) -> None:
+        # Patch any orphaned tool calls before clearing
+        from tunacode.core.agents.main import patch_tool_messages
+        patch_tool_messages("Conversation cleared", context.state_manager)
+        
         await ui.clear()
         context.state_manager.session.messages = []
+        await ui.success("Message history cleared")
+
+
+class FixCommand(SimpleCommand):
+    """Fix orphaned tool calls that cause API errors."""
+
+    def __init__(self):
+        super().__init__(
+            CommandSpec(
+                name="fix",
+                aliases=["/fix"],
+                description="Fix orphaned tool calls causing API errors",
+                category=CommandCategory.DEBUG,
+            )
+        )
+
+    async def execute(self, args: List[str], context: CommandContext) -> None:
+        from tunacode.core.agents.main import patch_tool_messages
+        
+        # Count current messages
+        before_count = len(context.state_manager.session.messages)
+        
+        # Patch orphaned tool calls
+        patch_tool_messages("Tool call resolved by /fix command", context.state_manager)
+        
+        # Count after patching
+        after_count = len(context.state_manager.session.messages)
+        patched_count = after_count - before_count
+        
+        if patched_count > 0:
+            await ui.success(f"Fixed {patched_count} orphaned tool call(s)")
+            await ui.muted("You can now continue the conversation normally")
+        else:
+            await ui.info("No orphaned tool calls found")
+
+
+class ParseToolsCommand(SimpleCommand):
+    """Parse and execute JSON tool calls from the last response."""
+
+    def __init__(self):
+        super().__init__(
+            CommandSpec(
+                name="parsetools",
+                aliases=["/parsetools"],
+                description="Parse JSON tool calls from last response when structured calling fails",
+                category=CommandCategory.DEBUG,
+            )
+        )
+
+    async def execute(self, args: List[str], context: CommandContext) -> None:
+        from tunacode.core.agents.main import extract_and_execute_tool_calls
+        
+        # Find the last model response in messages
+        messages = context.state_manager.session.messages
+        if not messages:
+            await ui.error("No message history found")
+            return
+        
+        # Look for the most recent response with text content
+        found_content = False
+        for msg in reversed(messages):
+            if hasattr(msg, 'parts'):
+                for part in msg.parts:
+                    if hasattr(part, 'content') and isinstance(part.content, str):
+                        # Create tool callback
+                        from tunacode.cli.repl import _tool_handler
+                        def tool_callback_with_state(part, node):
+                            return _tool_handler(part, node, context.state_manager)
+                        
+                        try:
+                            await extract_and_execute_tool_calls(
+                                part.content, 
+                                tool_callback_with_state, 
+                                context.state_manager
+                            )
+                            await ui.success("JSON tool parsing completed")
+                            found_content = True
+                            return
+                        except Exception as e:
+                            await ui.error(f"Failed to parse tools: {str(e)}")
+                            return
+        
+        if not found_content:
+            await ui.error("No parseable content found in recent messages")
+
+
+class RefreshConfigCommand(SimpleCommand):
+    """Refresh configuration from defaults."""
+
+    def __init__(self):
+        super().__init__(
+            CommandSpec(
+                name="refresh",
+                aliases=["/refresh"],
+                description="Refresh configuration from defaults (useful after updates)",
+                category=CommandCategory.SYSTEM,
+            )
+        )
+
+    async def execute(self, args: List[str], context: CommandContext) -> None:
+        from tunacode.configuration.defaults import DEFAULT_USER_CONFIG
+        
+        # Update current session config with latest defaults
+        for key, value in DEFAULT_USER_CONFIG.items():
+            if key not in context.state_manager.session.user_config:
+                context.state_manager.session.user_config[key] = value
+            elif isinstance(value, dict):
+                # Merge dict values, preserving user overrides
+                for subkey, subvalue in value.items():
+                    if subkey not in context.state_manager.session.user_config[key]:
+                        context.state_manager.session.user_config[key][subkey] = subvalue
+        
+        # Show updated max_iterations
+        max_iterations = context.state_manager.session.user_config.get("settings", {}).get("max_iterations", 20)
+        await ui.success(f"Configuration refreshed - max iterations: {max_iterations}")
 
 
 class TunaCodeCommand(SimpleCommand):
@@ -232,7 +418,6 @@ class HelpCommand(SimpleCommand):
         await ui.help(self._command_registry)
 
 
-
 class BranchCommand(SimpleCommand):
     """Create and switch to a new git branch."""
 
@@ -247,8 +432,8 @@ class BranchCommand(SimpleCommand):
         )
 
     async def execute(self, args: List[str], context: CommandContext) -> None:
-        import subprocess
         import os
+        import subprocess
 
         if not args:
             await ui.error("Usage: /branch <branch-name>")
@@ -332,14 +517,16 @@ class ModelCommand(SimpleCommand):
 
         # Get the model name from args
         model_name = args[0]
-        
+
         # Check if provider prefix is present
         if ":" not in model_name:
             await ui.error("Model name must include provider prefix")
             await ui.muted("Format: provider:model-name")
-            await ui.muted("Examples: openai:gpt-4.1, anthropic:claude-3-opus, google-gla:gemini-2.0-flash")
+            await ui.muted(
+                "Examples: openai:gpt-4.1, anthropic:claude-3-opus, google-gla:gemini-2.0-flash"
+            )
             return None
-        
+
         # No validation - user is responsible for correct model names
         await ui.warning("Model set without validation - verify the model name is correct")
 
@@ -416,8 +603,7 @@ class CommandRegistry:
         category_commands = self._categories[command.category]
         # Remove any existing instance of this command class
         self._categories[command.category] = [
-            cmd for cmd in category_commands 
-            if cmd.__class__ != command.__class__
+            cmd for cmd in category_commands if cmd.__class__ != command.__class__
         ]
         # Add the new instance
         self._categories[command.category].append(command)
@@ -436,7 +622,12 @@ class CommandRegistry:
         command_classes = [
             YoloCommand,
             DumpCommand,
+            ThoughtsCommand,
+            IterationsCommand,
             ClearCommand,
+            FixCommand,
+            ParseToolsCommand,
+            RefreshConfigCommand,
             HelpCommand,
             BranchCommand,
             # TunaCodeCommand,  # TODO: Temporarily disabled
@@ -459,7 +650,7 @@ class CommandRegistry:
         # Only update if callback has changed
         if self._factory.dependencies.process_request_callback == callback:
             return
-            
+
         self._factory.update_dependencies(process_request_callback=callback)
 
         # Re-register CompactCommand with new dependency if already registered
@@ -494,10 +685,10 @@ class CommandRegistry:
         if command_name in self._commands:
             command = self._commands[command_name]
             return await command.execute(args, context)
-        
+
         # Try partial matching
         matches = self.find_matching_commands(command_name)
-        
+
         if not matches:
             raise ValidationError(f"Unknown command: {command_name}")
         elif len(matches) == 1:
@@ -513,10 +704,10 @@ class CommandRegistry:
     def find_matching_commands(self, partial_command: str) -> List[str]:
         """
         Find all commands that start with the given partial command.
-        
+
         Args:
             partial_command: The partial command to match
-            
+
         Returns:
             List of matching command names
         """
@@ -534,11 +725,11 @@ class CommandRegistry:
             return False
 
         command_name = parts[0].lower()
-        
+
         # Check exact match first
         if command_name in self._commands:
             return True
-            
+
         # Check partial match
         return len(self.find_matching_commands(command_name)) > 0
 

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/src/tunacode/cli/repl.py

@@ -167,10 +167,14 @@ async def process_request(text: str, state_manager: StateManager, output: bool =
             await ui.error(str(e))
             return
 
+        # Patch any orphaned tool calls from previous requests before proceeding
+        patch_tool_messages("Tool execution was interrupted", state_manager)
+        
         # Create a partial function that includes state_manager
         def tool_callback_with_state(part, node):
             return _tool_handler(part, node, state_manager)
 
+        start_idx = len(state_manager.session.messages)
         res = await agent.process_request(
             state_manager.session.current_model,
             text,
@@ -178,7 +182,17 @@ async def process_request(text: str, state_manager: StateManager, output: bool =
             tool_callback=tool_callback_with_state,
         )
         if output:
-            await ui.agent(res.result.output)
+            if state_manager.session.show_thoughts:
+                new_msgs = state_manager.session.messages[start_idx:]
+                for msg in new_msgs:
+                    if isinstance(msg, dict) and "thought" in msg:
+                        await ui.muted(f"THOUGHT: {msg['thought']}")
+            # Check if result exists and has output
+            if hasattr(res, 'result') and res.result is not None and hasattr(res.result, 'output'):
+                await ui.agent(res.result.output)
+            else:
+                # Fallback: show that the request was processed
+                await ui.muted("Request completed")
     except CancelledError:
         await ui.muted("Request cancelled")
     except UserAbortError:
@@ -188,6 +202,31 @@ async def process_request(text: str, state_manager: StateManager, output: bool =
         await ui.muted(error_message)
         patch_tool_messages(error_message, state_manager)
     except Exception as e:
+        # Check if this might be a tool calling failure that we can recover from
+        error_str = str(e).lower()
+        if any(keyword in error_str for keyword in ['tool', 'function', 'call', 'schema']):
+            # Try to extract and execute tool calls from the last response
+            if state_manager.session.messages:
+                last_msg = state_manager.session.messages[-1]
+                if hasattr(last_msg, 'parts'):
+                    for part in last_msg.parts:
+                        if hasattr(part, 'content') and isinstance(part.content, str):
+                            from tunacode.core.agents.main import extract_and_execute_tool_calls
+                            try:
+                                # Create a partial function that includes state_manager
+                                def tool_callback_with_state(part, node):
+                                    return _tool_handler(part, node, state_manager)
+                                    
+                                await extract_and_execute_tool_calls(
+                                    part.content, 
+                                    tool_callback_with_state, 
+                                    state_manager
+                                )
+                                await ui.warning("🔧 Recovered using JSON tool parsing")
+                                return  # Successfully recovered
+                            except Exception:
+                                pass  # Fallback failed, continue with normal error handling
+        
         # Wrap unexpected exceptions in AgentError for better tracking
         agent_error = AgentError(f"Agent processing failed: {str(e)}")
         agent_error.__cause__ = e  # Preserve the original exception chain
@@ -210,7 +249,7 @@ async def repl(state_manager: StateManager):
     await ui.muted(f"• Model: {state_manager.session.current_model}")
     await ui.success("Ready to assist with your development")
     await ui.line()
-    
+
     instance = agent.get_or_create_agent(state_manager.session.current_model, state_manager)
 
     async with instance.run_mcp_servers():

{tunacode_cli-0.0.13 → tunacode_cli-0.0.14}/src/tunacode/configuration/defaults.py

@@ -18,6 +18,7 @@ DEFAULT_USER_CONFIG: UserConfig = {
     },
     "settings": {
         "max_retries": 10,
+        "max_iterations": 20,
         "tool_ignore": [TOOL_READ_FILE],
         "guide_file": GUIDE_FILE_NAME,
     },