jarvis-ai-assistant 0.1.102__tar.gz → 0.1.103__tar.gz

{jarvis_ai_assistant-0.1.102/src/jarvis_ai_assistant.egg-info → jarvis_ai_assistant-0.1.103}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: jarvis-ai-assistant
-Version: 0.1.102
+Version: 0.1.103
 Summary: Jarvis: An AI assistant that uses tools to interact with the system
 Home-page: https://github.com/skyfireitdiy/Jarvis
 Author: skyfire
@@ -45,10 +45,20 @@ Requires-Dist: colorama>=0.4.6
 Requires-Dist: prompt_toolkit>=3.0.0
 Requires-Dist: openai>=1.20.0
 Requires-Dist: playwright>=1.41.1
-Requires-Dist: numpy>=1.24.0
+Requires-Dist: numpy>=1.17.4
+Requires-Dist: faiss-cpu>=1.8.0
+Requires-Dist: sentence-transformers>=2.2.2
 Requires-Dist: bs4>=0.0.1
+Requires-Dist: PyMuPDF>=1.21.0
+Requires-Dist: python-docx>=0.8.11
+Requires-Dist: tiktoken>=0.3.0
 Requires-Dist: tqdm>=4.65.0
+Requires-Dist: docx>=0.2.4
 Requires-Dist: yaspin>=2.5.0
+Requires-Dist: rich>=13.3.1
+Requires-Dist: pygments>=2.15.0
+Requires-Dist: fuzzywuzzy>=0.18.0
+Requires-Dist: python-Levenshtein>=0.26.1
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Requires-Dist: black; extra == "dev"
@@ -148,6 +158,27 @@ Jarvis supports configuration through environment variables that can be set in t
 jarvis
 ```
 
+### Codebase Search
+```bash
+# Generate codebase index
+jarvis-codebase generate
+
+# Search similar code
+jarvis-codebase search "your search query"
+
+# Ask questions about codebase
+jarvis-codebase ask "your question"
+```
+
+### Document Analysis (RAG)
+```bash
+# Build document index
+jarvis-rag --dir /path/to/documents --build
+
+# Search documents
+jarvis-rag --query "your search query"
+```
+
 ### Search Tool
 ```bash
 # Basic search
@@ -162,6 +193,19 @@ jarvis-search "your query" --max 3
 
 ## 🛠️ Tools
 
+
+### Built-in Tools
+
+| Tool | Description |
+|------|-------------|
+| execute_shell | Execute system commands and capture output |
+| file_operation | File operations (read/write/append/delete) |
+| generate_tool | AI-powered tool generation and integration |
+| methodology | Experience accumulation and methodology management |
+| create_sub_agent | Create specialized sub-agents for specific tasks |
+| coder | Automatic code modification and generation tool |
+| codebase | Codebase management and search tool |
+
 ### Tool Locations
 - Built-in tools: `src/jarvis/tools/`
 - User tools: `~/.jarvis/tools/`
@@ -174,21 +218,27 @@ jarvis-search "your query" --max 3
 - Automatic code generation and integration
 - Dynamic capability expansion through sub-agents
 - Automatic code modification with version control
+- Codebase indexing and semantic search
 
 #### 2. Methodology Learning
 - Automatic experience accumulation from interactions
 - Pattern recognition and methodology extraction
 - Continuous refinement through usage
 - Code modification history tracking
+- Codebase analysis and documentation generation
 
 #### 3. Adaptive Problem Solving
 - Context-aware sub-agent creation
 - Dynamic tool composition
 - Learning from execution feedback
+- Codebase-aware problem solving
 - Multi-model collaboration for complex tasks
 
 #### 4. Code Intelligence
+- Automatic codebase indexing
+- Semantic code search
 - Code modification with git integration
+- Code analysis and documentation
 - Multi-model code generation
 
 ## 🎯 Extending Jarvis

{jarvis_ai_assistant-0.1.102 → jarvis_ai_assistant-0.1.103}/README.md

@@ -88,6 +88,27 @@ Jarvis supports configuration through environment variables that can be set in t
 jarvis
 ```
 
+### Codebase Search
+```bash
+# Generate codebase index
+jarvis-codebase generate
+
+# Search similar code
+jarvis-codebase search "your search query"
+
+# Ask questions about codebase
+jarvis-codebase ask "your question"
+```
+
+### Document Analysis (RAG)
+```bash
+# Build document index
+jarvis-rag --dir /path/to/documents --build
+
+# Search documents
+jarvis-rag --query "your search query"
+```
+
 ### Search Tool
 ```bash
 # Basic search
@@ -102,6 +123,19 @@ jarvis-search "your query" --max 3
 
 ## 🛠️ Tools
 
+
+### Built-in Tools
+
+| Tool | Description |
+|------|-------------|
+| execute_shell | Execute system commands and capture output |
+| file_operation | File operations (read/write/append/delete) |
+| generate_tool | AI-powered tool generation and integration |
+| methodology | Experience accumulation and methodology management |
+| create_sub_agent | Create specialized sub-agents for specific tasks |
+| coder | Automatic code modification and generation tool |
+| codebase | Codebase management and search tool |
+
 ### Tool Locations
 - Built-in tools: `src/jarvis/tools/`
 - User tools: `~/.jarvis/tools/`
@@ -114,21 +148,27 @@ jarvis-search "your query" --max 3
 - Automatic code generation and integration
 - Dynamic capability expansion through sub-agents
 - Automatic code modification with version control
+- Codebase indexing and semantic search
 
 #### 2. Methodology Learning
 - Automatic experience accumulation from interactions
 - Pattern recognition and methodology extraction
 - Continuous refinement through usage
 - Code modification history tracking
+- Codebase analysis and documentation generation
 
 #### 3. Adaptive Problem Solving
 - Context-aware sub-agent creation
 - Dynamic tool composition
 - Learning from execution feedback
+- Codebase-aware problem solving
 - Multi-model collaboration for complex tasks
 
 #### 4. Code Intelligence
+- Automatic codebase indexing
+- Semantic code search
 - Code modification with git integration
+- Code analysis and documentation
 - Multi-model code generation
 
 ## 🎯 Extending Jarvis

{jarvis_ai_assistant-0.1.102 → jarvis_ai_assistant-0.1.103}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "jarvis-ai-assistant"
-version = "0.1.102"
+version = "0.1.103"
 description = "Jarvis: An AI assistant that uses tools to interact with the system"
 readme = "README.md"
 authors = [{ name = "Your Name", email = "your.email@example.com" }]
@@ -27,10 +27,20 @@ dependencies = [
     "prompt_toolkit>=3.0.0",
     "openai>=1.20.0",
     "playwright>=1.41.1",
-    "numpy>=1.24.0",
+    "numpy>=1.17.4",
+    "faiss-cpu>=1.8.0",
+    "sentence-transformers>=2.2.2",
     "bs4>=0.0.1",
+    "PyMuPDF>=1.21.0",
+    "python-docx>=0.8.11",
+    "tiktoken>=0.3.0",
     "tqdm>=4.65.0",
+    "docx>=0.2.4",
     "yaspin>=2.5.0",
+    "rich>=13.3.1",
+    "pygments>=2.15.0",
+    "fuzzywuzzy>=0.18.0",
+    "python-Levenshtein>=0.26.1",
 ]
 requires-python = ">=3.8"
 
@@ -42,7 +52,12 @@ Homepage = "https://github.com/skyfireitdiy/Jarvis"
 
 [project.scripts]
 jarvis = "jarvis.agent:main"
-jarvis-code-agent = "jarvis.jarvis_code_agent.main:main"
+jarvis-code-agent = "jarvis.jarvis_code_agent.code_agent:main"
+jarvis-codebase = "jarvis.jarvis_codebase.main:main"
+jarvis-rag = "jarvis.jarvis_rag.main:main"
 jarvis-smart-shell = "jarvis.jarvis_smart_shell.main:main"
 jss = "jarvis.jarvis_smart_shell.main:main"
 jarvis-platform = "jarvis.jarvis_platform.main:main"
+jarvis-gh = "jarvis.jarvis_github.main:main"
+jarvis-git-commit = "jarvis.tools.git_commiter:main"
+jarvis-ctags = "jarvis.tools.create_ctags_agent:main"

{jarvis_ai_assistant-0.1.102 → jarvis_ai_assistant-0.1.103}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="jarvis-ai-assistant",
-    version="0.1.102",
+    version="0.1.103",
     author="skyfire",
     author_email="skyfireitdiy@hotmail.com",
     description="An AI assistant that uses various tools to interact with the system",
@@ -19,18 +19,33 @@ setup(
         "prompt_toolkit>=3.0.0",
         "openai>=1.20.0",
         "playwright>=1.41.1",
-        "numpy>=1.24.0",
+        "numpy==1.17.4",
+        "faiss-cpu>=1.8.0",
+        "sentence-transformers>=2.2.2",
         "bs4>=0.0.1",
+        "PyMuPDF>=1.21.0",
+        "python-docx>=0.8.11",
+        "tiktoken>=0.3.0",
         "tqdm>=4.65.0",
+        "docx>=0.2.4",
         "yaspin>=2.5.0",
+        "rich>=13.3.1",
+        "pygments>=2.15.0",
+        "fuzzywuzzy>=0.18.0",
+        "python-Levenshtein>=0.26.1",
     ],
     entry_points={
         "console_scripts": [
             "jarvis=jarvis.agent:main",
-            "jarvis-code-agent=jarvis.jarvis_code_agent.main:main",
+            "jarvis-code-agent=jarvis.jarvis_code_agent.code_agent:main",
+            "jarvis-codebase=jarvis.jarvis_codebase.main:main",
+            "jarvis-rag=jarvis.jarvis_rag.main:main",
             "jarvis-smart-shell=jarvis.jarvis_smart_shell.main:main",
             "jss=jarvis.jarvis_smart_shell.main:main",
             "jarvis-platform=jarvis.jarvis_platform.main:main",
+            "jarvis-gh=jarvis.jarvis_github.main:main",
+            "jarvis-git-commit=jarvis.tools.git_commiter:main",
+            "jarvis-ctags=jarvis.tools.create_ctags_agent:main",
         ],
     },
     python_requires=">=3.8",

{jarvis_ai_assistant-0.1.102 → jarvis_ai_assistant-0.1.103}/src/jarvis/__init__.py

@@ -1,3 +1,3 @@
 """Jarvis AI Assistant"""
 
-__version__ = "0.1.102" 
+__version__ = "0.1.103" 

{jarvis_ai_assistant-0.1.102 → jarvis_ai_assistant-0.1.103}/src/jarvis/agent.py

@@ -1,22 +1,38 @@
 import argparse
 import time
-from typing import Dict, List, Optional
+from typing import Callable, Dict, List, Optional
 
 from prompt_toolkit import prompt
 import yaml
 
+from jarvis.models.base import BasePlatform
 from jarvis.models.registry import PlatformRegistry
-from jarvis.tools import ToolRegistry
-from jarvis.tools.registry import load_tools
-from jarvis.utils import PrettyOutput, OutputType, get_single_line_input, load_methodology, add_agent, delete_current_agent, get_max_context_length, get_multiline_input, init_env
+from jarvis.tools.registry import ToolRegistry, tool_call_help
+from jarvis.utils import PrettyOutput, OutputType, is_auto_complete, load_methodology, add_agent, delete_current_agent, get_max_context_length, get_multiline_input, init_env
 import os
 
 class Agent:
 
     def __del__(self):
         delete_current_agent()
+
+    def set_summary_prompt(self, summary_prompt: str):
+        self.summary_prompt = summary_prompt
+
+    def set_output_filter(self, output_filter: List[Callable]):
+        self.output_filter = output_filter
         
-    def __init__(self, system_prompt: str, name: str = "Jarvis", is_sub_agent: bool = False, tool_registry: Optional[ToolRegistry] = None):
+    def __init__(self, 
+                 system_prompt: str, 
+                 name: str = "Jarvis", 
+                 is_sub_agent: bool = False, 
+                 tool_registry: Optional[ToolRegistry] = None, 
+                 platform: Optional[BasePlatform] = None, 
+                 summary_prompt: Optional[str] = None, 
+                 auto_complete: bool = False, 
+                 record_methodology: bool = True,
+                 output_filter: Optional[List[Callable]] = None,
+                 need_summary: bool = True):
         """Initialize Agent with a model, optional tool registry and name
         
         Args:
@@ -24,25 +40,70 @@ class Agent:
             name: Agent name, default is "Jarvis"
             is_sub_agent: Whether it is a sub-agent, default is False
             tool_registry: Tool registry instance
+            platform: Optional platform instance, default uses normal platform
         """
         add_agent(name)
         PrettyOutput.print(f"Welcome to Jarvis, your AI assistant, Initiating...", OutputType.SYSTEM)
-        self.model = PlatformRegistry.get_global_platform_registry().get_normal_platform()
-        self.tool_registry = tool_registry if tool_registry else ToolRegistry.get_global_tool_registry()
+        if platform is not None:
+            self.model = platform
+        else:
+            self.model = PlatformRegistry.get_global_platform_registry().get_normal_platform()
+        self.tool_registry = tool_registry if tool_registry else ToolRegistry()
+        self.record_methodology = record_methodology
         self.name = name
         self.is_sub_agent = is_sub_agent
         self.prompt = ""
         self.conversation_length = 0  # Use length counter instead
         self.system_prompt = system_prompt
+        self.need_summary = need_summary
         # Load configuration from environment variables
+        self.output_filter = output_filter if output_filter else []
+
+        self.summary_prompt = summary_prompt if summary_prompt else f"""Please generate a concise summary report of the task execution, including:
+
+1. Task Objective: Task restatement
+2. Execution Result: Success/Failure
+3. Key Information: Important information extracted during execution
+4. Important Findings: Any noteworthy discoveries
+5. Follow-up Suggestions: If any
+
+Please describe in concise bullet points, highlighting important information.
+"""
         
         self.max_context_length = get_max_context_length()
+
+        self.auto_complete = auto_complete
+
+
         
             
         # Initialize methodology related attributes
         self.methodology_data = []
 
         PrettyOutput.section(f"Jarvis initialized - With {self.model.name()}", OutputType.SYSTEM)
+        tools = self.tool_registry.get_all_tools()
+        if tools:
+            PrettyOutput.section(f"Available tools: {', '.join([tool['name'] for tool in tools])}", OutputType.SYSTEM)
+
+
+        # Load methodology
+        
+        tools_prompt = self.tool_registry.load_tools()
+        complete_prompt = """"""
+        if self.auto_complete:
+            complete_prompt = """
+            When the task is completed, you should print the following message:
+            <!!!COMPLETE!!!>
+            """
+
+        self.model.set_system_message(f"""
+{self.system_prompt}
+
+{tools_prompt}
+
+{complete_prompt}
+""")
+        self.first = True
 
     @staticmethod
     def extract_tool_calls(content: str) -> List[Dict]:
@@ -52,14 +113,6 @@ class Agent:
         tool_call_lines = []
         in_tool_call = False
 
-        tool_call_help = """Tool Usage Format:
-
-<TOOL_CALL>
-name: tool_name
-arguments:
-    param1: value1
-    param2: value2
-</TOOL_CALL>"""
 
         # Process line by line
         for line in lines:
@@ -68,27 +121,19 @@ arguments:
                 continue
             elif '</TOOL_CALL>' in line:
                 if in_tool_call and tool_call_lines:
-                    try:
-                        # Parse YAML directly
-                        tool_call_text = '\n'.join(tool_call_lines)
-                        tool_call_data = yaml.safe_load(tool_call_text)
-                        
-                        # Validate necessary fields
-                        if "name" in tool_call_data and "arguments" in tool_call_data:
-                            # Return content before tool call and tool call
-                            return [{
-                                "name": tool_call_data["name"],
-                                "arguments": tool_call_data["arguments"]
-                            }]
-                        else:
-                            PrettyOutput.print("Tool call missing necessary fields", OutputType.ERROR)
-                            raise Exception("Tool call missing necessary fields, " + tool_call_help)
-                    except yaml.YAMLError as e:
-                        PrettyOutput.print(f"YAML parsing error: {str(e)}", OutputType.ERROR)
-                        raise Exception(f"YAML parsing error: {str(e)}")
-                    except Exception as e:
-                        PrettyOutput.print(f"Error processing tool call: {str(e)}", OutputType.ERROR)
-                        raise Exception(f"Error processing tool call: {str(e)}")
+                    # Parse YAML directly
+                    tool_call_text = '\n'.join(tool_call_lines)
+                    tool_call_data = yaml.safe_load(tool_call_text)
+                    
+                    # Validate necessary fields
+                    if "name" in tool_call_data and "arguments" in tool_call_data:
+                        # Return content before tool call and tool call
+                        return [{
+                            "name": tool_call_data["name"],
+                            "arguments": tool_call_data["arguments"]
+                        }]
+                    else:
+                        raise Exception("Tool call missing necessary fields")
                 in_tool_call = False
                 continue
             
@@ -165,53 +210,39 @@ Please continue the task based on the above information.
         """
         PrettyOutput.section("Task completed", OutputType.SUCCESS)
         
-        # 询问是否生成方法论，带输入验证
-        while True:
-            user_input = get_single_line_input("Generate methodology for this task? (y/n)").strip().lower()
-            if user_input in ['y', 'n', '']:
-                break
-            PrettyOutput.print("Invalid input, please enter y or n", OutputType.WARNING)
-        
-        if user_input == 'y':
-            try:
-                # 让模型判断是否需要生成方法论
-                analysis_prompt = """The current task has ended, please analyze whether a methodology needs to be generated.
-If you think a methodology should be generated, first determine whether to create a new methodology or update an existing one. If updating an existing methodology, use 'update', otherwise use 'add'.
-If you think a methodology is not needed, please explain why.
-The methodology should be applicable to general scenarios, do not include task-specific information such as code commit messages.
-The methodology should include: problem restatement, optimal solution, notes (as needed), and nothing else.
-Only output the methodology tool call instruction, or the explanation for not generating a methodology. Do not output anything else.
-"""
-                self.prompt = analysis_prompt
-                response = self._call_model(self.prompt)
-                
-                # 检查是否包含工具调用
+        if not self.is_sub_agent:
+            if self.record_methodology:
+
                 try:
-                    tool_calls = Agent.extract_tool_calls(response)
-                    if tool_calls:
-                        self.tool_registry.handle_tool_calls(tool_calls)
+                    # 让模型判断是否需要生成方法论
+                    analysis_prompt = """The current task has ended, please analyze whether a methodology needs to be generated.
+    If you think a methodology should be generated, first determine whether to create a new methodology or update an existing one. If updating an existing methodology, use 'update', otherwise use 'add'.
+    If you think a methodology is not needed, please explain why.
+    The methodology should be applicable to general scenarios, do not include task-specific information such as code commit messages.
+    The methodology should include: problem restatement, optimal solution, notes (as needed), and nothing else.
+    Only output the methodology tool call instruction, or the explanation for not generating a methodology. Do not output anything else.
+    """
+                    self.prompt = analysis_prompt
+                    response = self._call_model(self.prompt)
+                    
+                    # 检查是否包含工具调用
+                    try:
+                        tool_calls = Agent.extract_tool_calls(response)
+                        if tool_calls:
+                            self.tool_registry.handle_tool_calls(tool_calls)
+                    except Exception as e:
+                        PrettyOutput.print(f"Failed to handle methodology generation: {str(e)}", OutputType.ERROR)
+                    
                 except Exception as e:
-                    PrettyOutput.print(f"Failed to handle methodology generation: {str(e)}", OutputType.ERROR)
-                
-            except Exception as e:
-                PrettyOutput.print(f"Error generating methodology: {str(e)}", OutputType.ERROR)
-        
-        if not self.is_sub_agent:
+                    PrettyOutput.print(f"Error generating methodology: {str(e)}", OutputType.ERROR)
+            
             return "Task completed"
         
-        # 生成任务总结
-        summary_prompt = f"""Please generate a concise summary report of the task execution, including:
-
-1. Task Objective: Task restatement
-2. Execution Result: Success/Failure
-3. Key Information: Important information extracted during execution
-4. Important Findings: Any noteworthy discoveries
-5. Follow-up Suggestions: If any
-
-Please describe in concise bullet points, highlighting important information.
-"""
-        self.prompt = summary_prompt
-        return self._call_model(self.prompt)
+        if self.need_summary:
+            self.prompt = self.summary_prompt
+            return self._call_model(self.prompt)
+        
+        return "Task completed"
 
 
     def run(self, user_input: str, file_list: Optional[List[str]] = None) -> str:
@@ -224,28 +255,22 @@ Please describe in concise bullet points, highlighting important information.
         Returns:
             str: Task summary report
         """
+
+        
+
         try:
             PrettyOutput.section("Preparing environment", OutputType.PLANNING)
             if file_list:
                 self.model.upload_files(file_list)
 
-            # Load methodology
-            methodology_prompt = load_methodology(user_input)
-            tools_prompt = load_tools()
-
             # 显示任务开始
             PrettyOutput.section(f"Starting new task: {self.name}", OutputType.PLANNING)
 
-            self.clear_history()  
-
-            self.model.set_system_message(f"""
-{self.system_prompt}
-
-{tools_prompt}
-
-{methodology_prompt}
-""")
-            self.prompt = f"{user_input}"
+            if self.first:
+                self.prompt = f"{user_input}\n\n{load_methodology(user_input)}"
+                self.first = False
+            else:
+                self.prompt = f"{user_input}"
 
             while True:
                 try:
@@ -261,25 +286,32 @@ Please describe in concise bullet points, highlighting important information.
                         continue
                     else:
                         current_response = self._call_model(self.prompt)
-                        self.conversation_length += len(current_response)  # Add response length
+                        self.prompt = ""
+                        self.conversation_length += len(current_response) 
+
+                    for filter in self.output_filter:
+                        self.prompt += filter(current_response)
+
                     try:
                         result = Agent.extract_tool_calls(current_response)
                     except Exception as e:
                         PrettyOutput.print(f"Tool call error: {str(e)}", OutputType.ERROR)
-                        self.prompt = f"Tool call error: {str(e)}"
+                        self.prompt += f"Tool call error: {str(e)}"
                         continue
                     
                     if len(result) > 0:
                         PrettyOutput.print("Executing tool call...", OutputType.PROGRESS)
                         tool_result = self.tool_registry.handle_tool_calls(result)
-                        self.prompt = tool_result
+                        self.prompt += tool_result
+                    
+                    if self.prompt:
                         continue
+
+                    if self.auto_complete and "<!!!COMPLETE!!!>" in current_response:
+                        return self._complete_task()
                     
                     # 获取用户输入
                     user_input = get_multiline_input(f"{self.name}: You can continue to input, or enter an empty line to end the current task")
-                    if user_input == "__interrupt__":
-                        PrettyOutput.print("Task cancelled by user", OutputType.WARNING)
-                        return "Task cancelled by user"
 
                     if user_input:
                         self.prompt = user_input
@@ -362,10 +394,11 @@ def select_task(tasks: dict) -> str:
     # Convert tasks to list for ordered display
     task_names = list(tasks.keys())
     
-    PrettyOutput.print("\nAvailable tasks:", OutputType.INFO)
+    task_list = ["Available tasks:"]
     for i, name in enumerate(task_names, 1):
-        PrettyOutput.print(f"[{i}] {name}", OutputType.INFO)
-    PrettyOutput.print("[0] Skip predefined tasks", OutputType.INFO)
+        task_list.append(f"[{i}] {name}")
+    task_list.append("[0] Skip predefined tasks")
+    PrettyOutput.print("\n".join(task_list), OutputType.INFO)
     
     
     while True:
@@ -406,25 +439,13 @@ When users need to execute tasks, you will strictly follow these steps to handle
 7. Execute Action Plan: Execute one step at a time, **use at most one tool** (wait for tool execution results before proceeding)
 8. Monitor and Adjust: If execution results don't match expectations, reflect and adjust the action plan, iterate previous steps
 9. Methodology: If the current task has general applicability and valuable experience is gained, use methodology tools to record it for future similar problems
-10. Task Completion: End the task using task completion command when finished
+10. Auto check the task goal completion status: If the task goal is completed, use the task completion command to end the task
+11. Task Completion: End the task using task completion command when finished
 
 Methodology Template:
 1. Problem Restatement
 2. Optimal Solution
 3. Optimal Solution Steps (exclude failed actions)
-                                          
-Strict Rules:
-- Execute only one tool at a time
-- Tool execution must strictly follow the tool usage format
-- Wait for user to provide execution results
-- Don't assume or imagine results
-- Don't create fake dialogues
-- If current information is insufficient, you may ask the user
-- Not all problem-solving steps are mandatory, skip as appropriate
-- Request user guidance when multiple iterations show no progress
-- If yaml string contains colons, wrap the entire string in quotes to avoid yaml parsing errors
-- Use | syntax for multi-line strings in yaml
-- If you can start executing the task, please start directly without asking the user if you can begin.
 
 -------------------------------------------------------------"""
 
@@ -453,7 +474,7 @@ def main():
         while True:
             try:
                 user_input = get_multiline_input("Please enter your task (input empty line to exit):")
-                if not user_input or user_input == "__interrupt__":
+                if not user_input:
                     break
                 agent.run(user_input, args.files)
             except Exception as e: