devduck 1.1.0__tar.gz → 1.1.4__tar.gz

{devduck-1.1.0 → devduck-1.1.4}/.github/workflows/agent.yml

@@ -1,8 +1,8 @@
 name: DevDuck
 
 on:
-  schedule:
-    - cron: '0 8 * * *'
+  # schedule:
+  #   - cron: '0 8 * * *'
   issues:
     types: [opened, edited, closed, reopened, assigned, unassigned, labeled, unlabeled]
   issue_comment:

{devduck-1.1.0 → devduck-1.1.4}/.gitignore

@@ -7,4 +7,7 @@ dist/
 build
 _version.py
 .bedrock_agentcore.yaml
-.bedrock_agentcore
+.bedrock_agentcore
+google_*.json
+gmail_*.json
+.DS_Store

{devduck-1.1.0 → devduck-1.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: devduck
-Version: 1.1.0
+Version: 1.1.4
 Summary: 🦆 Extreme minimalist self-adapting AI agent - one file, self-healing, runtime dependencies
 Author-email: Cagatay Cali <cagataycali@icloud.com>
 License: Apache-2.0
@@ -40,6 +40,7 @@ Requires-Dist: beautifulsoup4
 Requires-Dist: colorama
 Requires-Dist: websockets
 Requires-Dist: strands-mcp-server
+Requires-Dist: strands-google
 Requires-Dist: bedrock-agentcore-starter-toolkit
 Requires-Dist: bedrock-agentcore
 Requires-Dist: rumps; sys_platform == "darwin"

{devduck-1.1.0 → devduck-1.1.4}/devduck/__init__.py

@@ -3,19 +3,31 @@
 🦆 devduck - extreme minimalist self-adapting agent
 one file. self-healing. runtime dependencies. adaptive.
 """
+import os
 import sys
 import subprocess
-import os
+import threading
 import platform
 import socket
 import logging
 import tempfile
 import time
 import warnings
+import json
 from pathlib import Path
 from datetime import datetime
 from typing import Dict, Any
 from logging.handlers import RotatingFileHandler
+from strands import Agent, tool
+
+# Import system prompt helper for loading prompts from files
+try:
+    from devduck.tools.system_prompt import _get_system_prompt
+except ImportError:
+    # Fallback if tools module not available yet
+    def _get_system_prompt(repository=None, variable_name="SYSTEM_PROMPT"):
+        return os.getenv(variable_name, "")
+
 
 warnings.filterwarnings("ignore", message=".*pkg_resources is deprecated.*")
 warnings.filterwarnings("ignore", message=".*cache_prompt is deprecated.*")
@@ -594,7 +606,7 @@ class DevDuck:
                 servers = {
                     "tcp": {
                         "port": int(os.getenv("DEVDUCK_TCP_PORT", "9999")),
-                        "enabled": os.getenv("DEVDUCK_ENABLE_TCP", "true").lower()
+                        "enabled": os.getenv("DEVDUCK_ENABLE_TCP", "false").lower()
                         == "true",
                     },
                     "ws": {
@@ -604,25 +616,93 @@ class DevDuck:
                     },
                     "mcp": {
                         "port": int(os.getenv("DEVDUCK_MCP_PORT", "8000")),
-                        "enabled": os.getenv("DEVDUCK_ENABLE_MCP", "true").lower()
+                        "enabled": os.getenv("DEVDUCK_ENABLE_MCP", "false").lower()
                         == "true",
                     },
                     "ipc": {
                         "socket_path": os.getenv(
                             "DEVDUCK_IPC_SOCKET", "/tmp/devduck_main.sock"
                         ),
-                        "enabled": os.getenv("DEVDUCK_ENABLE_IPC", "true").lower()
+                        "enabled": os.getenv("DEVDUCK_ENABLE_IPC", "false").lower()
                         == "true",
                     },
                 }
 
-            self.servers = servers
+            # Show server configuration status
+            enabled_servers = []
+            disabled_servers = []
+            for server_name, config in servers.items():
+                if config.get("enabled", False):
+                    if "port" in config:
+                        enabled_servers.append(
+                            f"{server_name.upper()}:{config['port']}"
+                        )
+                    else:
+                        enabled_servers.append(server_name.upper())
+                else:
+                    disabled_servers.append(server_name.upper())
 
-            from strands import Agent, tool
+            logger.debug(
+                f"🦆 Server config: {', '.join(enabled_servers) if enabled_servers else 'none enabled'}"
+            )
+            if disabled_servers:
+                logger.debug(f"🦆 Disabled: {', '.join(disabled_servers)}")
+
+            self.servers = servers
 
             # Load tools with flexible configuration
-            # Default tool config - user can override with DEVDUCK_TOOLS env var
-            default_tools = "devduck.tools:system_prompt,store_in_kb,ipc,tcp,websocket,mcp_server,state_manager,tray,ambient,agentcore_config,agentcore_invoke,agentcore_logs,agentcore_agents,install_tools,create_subagent,use_github;strands_tools:shell,editor,file_read,file_write,image_reader,load_tool,retrieve,calculator,use_agent,environment,mcp_client,speak,slack;strands_fun_tools:listen,cursor,clipboard,screen_reader,bluetooth,yolo_vision"
+            # Default tool config
+            # Agent can load additional tools on-demand via fetch_github_tool
+
+            # 🔧 Available DevDuck Tools (load on-demand):
+            # - system_prompt: https://github.com/cagataycali/devduck/blob/main/devduck/tools/system_prompt.py
+            # - store_in_kb: https://github.com/cagataycali/devduck/blob/main/devduck/tools/store_in_kb.py
+            # - ipc: https://github.com/cagataycali/devduck/blob/main/devduck/tools/ipc.py
+            # - tcp: https://github.com/cagataycali/devduck/blob/main/devduck/tools/tcp.py
+            # - websocket: https://github.com/cagataycali/devduck/blob/main/devduck/tools/websocket.py
+            # - mcp_server: https://github.com/cagataycali/devduck/blob/main/devduck/tools/mcp_server.py
+            # - scraper: https://github.com/cagataycali/devduck/blob/main/devduck/tools/scraper.py
+            # - tray: https://github.com/cagataycali/devduck/blob/main/devduck/tools/tray.py
+            # - ambient: https://github.com/cagataycali/devduck/blob/main/devduck/tools/ambient.py
+            # - agentcore_config: https://github.com/cagataycali/devduck/blob/main/devduck/tools/agentcore_config.py
+            # - agentcore_invoke: https://github.com/cagataycali/devduck/blob/main/devduck/tools/agentcore_invoke.py
+            # - agentcore_logs: https://github.com/cagataycali/devduck/blob/main/devduck/tools/agentcore_logs.py
+            # - agentcore_agents: https://github.com/cagataycali/devduck/blob/main/devduck/tools/agentcore_agents.py
+            # - create_subagent: https://github.com/cagataycali/devduck/blob/main/devduck/tools/create_subagent.py
+            # - use_github: https://github.com/cagataycali/devduck/blob/main/devduck/tools/use_github.py
+            # - speech_to_speech: https://github.com/cagataycali/devduck/blob/main/devduck/tools/speech_to_speech.py
+            # - state_manager: https://github.com/cagataycali/devduck/blob/main/devduck/tools/state_manager.py
+
+            # 📦 Strands Tools
+            # - editor, file_read, file_write, image_reader, load_tool, retrieve
+            # - calculator, use_agent, environment, mcp_client, speak, slack
+
+            # 🎮 Strands Fun Tools
+            # - listen, cursor, clipboard, screen_reader, bluetooth, yolo_vision
+
+            # 🔍 Strands Google
+            # - use_google, google_auth
+
+            # 🔧 Auto-append server tools based on enabled servers
+            server_tools_needed = []
+            if servers.get("tcp", {}).get("enabled", False):
+                server_tools_needed.append("tcp")
+            if servers.get("ws", {}).get("enabled", True):
+                server_tools_needed.append("websocket")
+            if servers.get("mcp", {}).get("enabled", False):
+                server_tools_needed.append("mcp_server")
+            if servers.get("ipc", {}).get("enabled", False):
+                server_tools_needed.append("ipc")
+
+            # Append to default tools if any server tools are needed
+            if server_tools_needed:
+                server_tools_str = ",".join(server_tools_needed)
+                default_tools = f"devduck.tools:system_prompt,fetch_github_tool,{server_tools_str};strands_tools:shell"
+                logger.info(f"Auto-added server tools: {server_tools_str}")
+            else:
+                default_tools = (
+                    "devduck.tools:system_prompt,fetch_github_tool;strands_tools:shell"
+                )
 
             tools_config = os.getenv("DEVDUCK_TOOLS", default_tools)
             logger.info(f"Loading tools from config: {tools_config}")
@@ -776,8 +856,6 @@ class DevDuck:
         Returns:
             List of MCPClient instances ready for direct use in Agent
         """
-        import json
-
         mcp_servers_json = os.getenv("MCP_SERVERS")
         if not mcp_servers_json:
             logger.debug("No MCP_SERVERS environment variable found")
@@ -1145,7 +1223,7 @@ When you learn something valuable during conversations:
 - Communication: **MINIMAL WORDS**
 - Efficiency: **Speed is paramount**
 
-{os.getenv('SYSTEM_PROMPT', '')}"""
+{_get_system_prompt()}"""
 
     def _self_heal(self, error):
         """Attempt self-healing when errors occur"""
@@ -1192,7 +1270,6 @@ When you learn something valuable during conversations:
 
     def _is_socket_available(self, socket_path):
         """Check if a Unix socket is available"""
-        import os
 
         # If socket file doesn't exist, it's available
         if not os.path.exists(socket_path):
@@ -1430,11 +1507,11 @@ When you learn something valuable during conversations:
     def restart(self):
         """Restart the agent"""
         print("\n🦆 Restarting...")
+        logger.debug("\n🦆 Restarting...")
         self.__init__()
 
     def _start_file_watcher(self):
         """Start background file watcher for auto hot-reload"""
-        import threading
 
         logger.info("Starting file watcher for hot-reload")
         # Get the path to this file
@@ -1534,6 +1611,7 @@ When you learn something valuable during conversations:
                 self._watcher_running = False
 
             print("\n🦆 Restarting process with fresh code...")
+            logger.debug("\n🦆 Restarting process with fresh code...")
 
             # Restart the entire Python process
             # This ensures all code is freshly loaded
@@ -1677,7 +1755,7 @@ def interactive():
     print(f"📝 Logs: {LOG_DIR}")
     print("Type 'exit', 'quit', or 'q' to quit.")
     print("Prefix with ! to run shell commands (e.g., ! ls -la)")
-    print("-" * 50)
+    print("\n\n")
     logger.info("Interactive mode started")
 
     # Set up prompt_toolkit with history
@@ -1705,7 +1783,6 @@ def interactive():
                 auto_suggest=AutoSuggestFromHistory(),
                 completer=completer,
                 complete_while_typing=True,
-                mouse_support=False,  # breaks scrolling when enabled
             )
 
             # Reset interrupt count on successful prompt
@@ -1733,6 +1810,10 @@ def interactive():
                         )
                         devduck._agent_executing = False
 
+                        # Reset terminal to fix rendering issues after command output
+                        print("\r", end="", flush=True)
+                        sys.stdout.flush()
+
                         # Append shell command to history
                         append_to_shell_history(q, result["content"][0]["text"])
 
@@ -1747,6 +1828,9 @@ def interactive():
                 except Exception as e:
                     devduck._agent_executing = False  # Reset on error
                     print(f"🦆 Shell command error: {e}")
+                    # Reset terminal on error too
+                    print("\r", end="", flush=True)
+                    sys.stdout.flush()
                 continue
 
             # Execute the agent with user input

{devduck-1.1.0 → devduck-1.1.4}/devduck/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '1.1.0'
-__version_tuple__ = version_tuple = (1, 1, 0)
+__version__ = version = '1.1.4'
+__version_tuple__ = version_tuple = (1, 1, 4)
 
-__commit_id__ = commit_id = 'g6a4263e8c'
+__commit_id__ = commit_id = 'g9882068fd'

devduck-1.1.4/devduck/tools/__init__.py

@@ -0,0 +1,47 @@
+"""
+DevDuck Tools Package
+
+This module exports all available tools for devduck.
+"""
+
+from .agentcore_agents import agentcore_agents
+from .agentcore_config import agentcore_config
+from .agentcore_invoke import agentcore_invoke
+from .agentcore_logs import agentcore_logs
+from .ambient import ambient
+from .create_subagent import create_subagent
+from .fetch_github_tool import fetch_github_tool
+from .install_tools import install_tools
+from .ipc import ipc
+from .mcp_server import mcp_server
+from .scraper import scraper
+from .speech_to_speech import speech_to_speech
+from .state_manager import state_manager
+from .store_in_kb import store_in_kb
+from .system_prompt import system_prompt
+from .tcp import tcp
+from .tray import tray
+from .use_github import use_github
+from .websocket import websocket
+
+__all__ = [
+    "agentcore_agents",
+    "agentcore_config",
+    "agentcore_invoke",
+    "agentcore_logs",
+    "ambient",
+    "create_subagent",
+    "fetch_github_tool",
+    "install_tools",
+    "ipc",
+    "mcp_server",
+    "scraper",
+    "speech_to_speech",
+    "state_manager",
+    "store_in_kb",
+    "system_prompt",
+    "tcp",
+    "tray",
+    "use_github",
+    "websocket",
+]

{devduck-1.1.0 → devduck-1.1.4}/devduck/tools/speech_to_speech.py

@@ -102,15 +102,21 @@ class SpeechSession:
         self,
         session_id: str,
         agent: BidiAgent,
+        input_device_index: Optional[int] = None,
+        output_device_index: Optional[int] = None,
     ):
         """Initialize speech session.
 
         Args:
             session_id: Unique session identifier
             agent: BidiAgent instance
+            input_device_index: PyAudio input device index
+            output_device_index: PyAudio output device index
         """
         self.session_id = session_id
         self.agent = agent
+        self.input_device_index = input_device_index
+        self.output_device_index = output_device_index
         self.active = False
         self.thread = None
         self.loop = None
@@ -187,8 +193,11 @@ class SpeechSession:
     async def _async_session(self) -> None:
         """Async session management using BidiAudioIO."""
         try:
-            # Create audio I/O
-            audio_io = BidiAudioIO()
+            # Create audio I/O with device indices
+            audio_io = BidiAudioIO(
+                input_device_index=self.input_device_index,
+                output_device_index=self.output_device_index,
+            )
 
             # Run agent with audio I/O
             await self.agent.run(inputs=[audio_io.input()], outputs=[audio_io.output()])
@@ -207,6 +216,9 @@ def speech_to_speech(
     tools: Optional[List[str]] = None,
     agent: Optional[Any] = None,
     load_history_from: Optional[str] = None,
+    inherit_system_prompt: bool = False,
+    input_device_index: Optional[int] = None,
+    output_device_index: Optional[int] = None,
 ) -> str:
     """Start, stop, or manage speech-to-speech conversations.
 
@@ -221,18 +233,22 @@ def speech_to_speech(
             - "status": Get session status
             - "list_history": List saved conversation histories
             - "read_history": Read a specific conversation history
+            - "list_audio_devices": List all available audio input/output devices
         provider: Model provider to use:
             - "novasonic": AWS Bedrock Nova Sonic
             - "openai": OpenAI Realtime API
             - "gemini_live": Google Gemini Live
         system_prompt: Custom system prompt for the agent. This will be appended
-            to the parent agent's system prompt (if available). If not provided,
-            uses default prompt that encourages tool usage.
+            to the parent agent's system prompt (if inherit_system_prompt=True).
+            If not provided, uses default prompt that encourages tool usage.
         session_id: Session identifier:
             - For "start": Custom ID (auto-generated if not provided)
             - For "stop": Specific session to stop (stops all if not provided)
             - For "read_history": Session ID to read history from
             - For "status": Not used
+        inherit_system_prompt: Whether to inherit parent agent's system prompt.
+            Set to False to use only the custom system_prompt (useful for OpenAI
+            which has 16K token limit). Default: False
         model_settings: Provider-specific configuration dictionary. Structure:
             {
                 "model_id": "model-name",
@@ -260,6 +276,10 @@ def speech_to_speech(
         agent: Parent agent (automatically passed by Strands framework)
         load_history_from: Optional session ID to load conversation history from
             when starting a new session (provides context continuity)
+        input_device_index: Optional PyAudio input device index. If not specified,
+            uses system default. Use action="list_audio_devices" to see available devices.
+        output_device_index: Optional PyAudio output device index. If not specified,
+            uses system default. Use action="list_audio_devices" to see available devices.
 
     Returns:
         str: Status message with session details or error information
@@ -287,6 +307,9 @@ def speech_to_speech(
             tools,
             agent,
             load_history_from,
+            inherit_system_prompt,
+            input_device_index,
+            output_device_index,
         )
     elif action == "stop":
         return _stop_speech_session(session_id)
@@ -296,6 +319,8 @@ def speech_to_speech(
         return _list_conversation_histories()
     elif action == "read_history":
         return _read_conversation_history(session_id)
+    elif action == "list_audio_devices":
+        return _list_audio_devices()
     else:
         return f"Unknown action: {action}"
 
@@ -364,6 +389,9 @@ def _start_speech_session(
     tool_names: Optional[List[str]],
     parent_agent: Optional[Any],
     load_history_from: Optional[str],
+    inherit_system_prompt: bool,
+    input_device_index: Optional[int],
+    output_device_index: Optional[int],
 ) -> str:
     """Start a speech-to-speech session with full configuration support."""
     try:
@@ -384,7 +412,7 @@ def _start_speech_session(
             if provider == "novasonic":
                 # Nova Sonic only available in: us-east-1, eu-north-1, ap-northeast-1
                 default_settings = {
-                    "model_id": "amazon.nova-2-sonic-v1:0",
+                    "model_id": os.getenv("BIDI_MODEL_ID", "amazon.nova-2-sonic-v1:0"),
                     "provider_config": {
                         "audio": {
                             "voice": "tiffany",
@@ -421,7 +449,7 @@ def _start_speech_session(
             elif provider == "openai":
                 # Read API key from environment if not provided in model_settings
                 default_settings = {
-                    "model_id": "gpt-realtime",
+                    "model_id": os.getenv("BIDI_MODEL_ID", "gpt-realtime"),
                     "client_config": {
                         "api_key": os.getenv("OPENAI_API_KEY"),
                     },
@@ -457,7 +485,9 @@ def _start_speech_session(
                 api_key = os.getenv("GOOGLE_API_KEY") or os.getenv("GEMINI_API_KEY")
 
                 default_settings = {
-                    "model_id": "gemini-2.5-flash-native-audio-preview-09-2025",
+                    "model_id": os.getenv(
+                        "BIDI_MODEL_ID", "gemini-2.5-flash-native-audio-preview-09-2025"
+                    ),
                     "client_config": {
                         "api_key": api_key,
                     },
@@ -545,8 +575,12 @@ def _start_speech_session(
         # Build system prompt: parent prompt + custom prompt
         final_system_prompt = ""
 
-        # Get parent agent's system prompt if available
-        if parent_agent and hasattr(parent_agent, "system_prompt"):
+        # Get parent agent's system prompt if available and inheritance enabled
+        if (
+            inherit_system_prompt
+            and parent_agent
+            and hasattr(parent_agent, "system_prompt")
+        ):
             parent_prompt = parent_agent.system_prompt or ""
             if parent_prompt:
                 final_system_prompt = parent_prompt
@@ -581,6 +615,8 @@ Keep your voice responses brief and natural."""
         session = SpeechSession(
             session_id=session_id,
             agent=bidi_agent,
+            input_device_index=input_device_index,
+            output_device_index=output_device_index,
         )
 
         session.start()
@@ -748,3 +784,67 @@ def _read_conversation_history(session_id: Optional[str]) -> str:
 
     except Exception as e:
         return f"❌ Error reading history: {e}"
+
+
+def _list_audio_devices() -> str:
+    """List all available audio input and output devices."""
+    try:
+        import pyaudio
+
+        p = pyaudio.PyAudio()
+
+        lines = ["**Available Audio Devices:**\n"]
+
+        # List all devices
+        device_count = p.get_device_count()
+        default_input = p.get_default_input_device_info()["index"]
+        default_output = p.get_default_output_device_info()["index"]
+
+        lines.append(f"Total devices: {device_count}\n")
+
+        for i in range(device_count):
+            try:
+                info = p.get_device_info_by_index(i)
+                name = info["name"]
+                max_input_channels = info["maxInputChannels"]
+                max_output_channels = info["maxOutputChannels"]
+
+                device_type = []
+                is_default = []
+
+                if max_input_channels > 0:
+                    device_type.append("INPUT")
+                    if i == default_input:
+                        is_default.append("default input")
+
+                if max_output_channels > 0:
+                    device_type.append("OUTPUT")
+                    if i == default_output:
+                        is_default.append("default output")
+
+                type_str = "/".join(device_type) if device_type else "NONE"
+                default_str = f" [{', '.join(is_default)}]" if is_default else ""
+
+                lines.append(
+                    f"- **Index {i}:** {name}\n"
+                    f"  Type: {type_str}{default_str}\n"
+                    f"  Input Channels: {max_input_channels}, Output Channels: {max_output_channels}"
+                )
+
+            except Exception as e:
+                lines.append(f"- **Index {i}:** Error reading device info - {e}")
+
+        p.terminate()
+
+        lines.append(
+            "\n**Usage:**\n"
+            "To use a specific device, pass the index:\n"
+            '  speech_to_speech(action="start", input_device_index=2, output_device_index=5)'
+        )
+
+        return "\n".join(lines)
+
+    except ImportError:
+        return "❌ PyAudio not installed. Install with: pip install pyaudio"
+    except Exception as e:
+        return f"❌ Error listing audio devices: {e}"