code-puppy 0.0.171__tar.gz → 0.0.172__tar.gz

{code_puppy-0.0.171 → code_puppy-0.0.172}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code-puppy
-Version: 0.0.171
+Version: 0.0.172
 Summary: Code generation agent
 Project-URL: repository, https://github.com/mpfaffenberger/code_puppy
 Project-URL: HomePage, https://github.com/mpfaffenberger/code_puppy
@@ -16,6 +16,7 @@ Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Software Development :: Code Generators
 Requires-Python: >=3.11
 Requires-Dist: bs4>=0.0.2
+Requires-Dist: camoufox>=0.4.11
 Requires-Dist: fastapi>=0.110.0
 Requires-Dist: httpx-limiter>=0.3.0
 Requires-Dist: httpx>=0.24.1
@@ -23,6 +24,7 @@ Requires-Dist: json-repair>=0.46.2
 Requires-Dist: logfire>=0.7.1
 Requires-Dist: openai>=1.99.1
 Requires-Dist: pathspec>=0.11.0
+Requires-Dist: playwright>=1.40.0
 Requires-Dist: prompt-toolkit>=3.0.38
 Requires-Dist: pydantic-ai>=1.0.10
 Requires-Dist: pydantic>=2.4.0

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/agent.py

@@ -48,7 +48,7 @@ _code_generation_agent = None
 def _load_mcp_servers(extra_headers: Optional[Dict[str, str]] = None):
     """Load MCP servers using the new manager while maintaining backward compatibility."""
     from code_puppy.config import get_value, load_mcp_server_configs
-    from code_puppy.mcp import ServerConfig, get_mcp_manager
+    from code_puppy.mcp_ import ServerConfig, get_mcp_manager
 
     # Check if MCP servers are disabled
     mcp_disabled = get_value("disable_mcp_servers")
@@ -113,7 +113,7 @@ def _load_mcp_servers(extra_headers: Optional[Dict[str, str]] = None):
 
 def reload_mcp_servers():
     """Reload MCP servers without restarting the agent."""
-    from code_puppy.mcp import get_mcp_manager
+    from code_puppy.mcp_ import get_mcp_manager
 
     manager = get_mcp_manager()
     # Reload configurations
@@ -174,7 +174,7 @@ def reload_code_generation_agent(message_group: str | None):
 
     model_settings = ModelSettings(**model_settings_dict)
     if "gpt-5" in model_name:
-        model_settings_dict["openai_reasoning_effort"] = "high"
+        model_settings_dict["openai_reasoning_effort"] = "off"
         model_settings_dict["extra_body"] = {
             "verbosity": "low"
         }

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/agents/agent_creator_agent.py

@@ -245,9 +245,6 @@ Best-practice guidelines for `invoke_agent`:
 Your solutions should be production-ready, maintainable, and follow best practices for the chosen language.
 
 Return your final response as a string output
-Walmart specific rules:
- - You are operating inside Walmart Global Tech! Yay!
- - Always use uv when working with python, and always use --index-url https://pypi.ci.artifacts.walmart.com/artifactory/api/pypi/external-pypi/simple
 
 ## Tool Templates:
 

code_puppy-0.0.172/code_puppy/agents/agent_qa_kitten.py

@@ -0,0 +1,203 @@
+"""Quality Assurance Kitten - Playwright-powered browser automation agent."""
+
+from .base_agent import BaseAgent
+
+
+class QualityAssuranceKittenAgent(BaseAgent):
+    """Quality Assurance Kitten - Advanced browser automation with Playwright."""
+
+    @property
+    def name(self) -> str:
+        return "qa-kitten"
+
+    @property
+    def display_name(self) -> str:
+        return "Quality Assurance Kitten 🐱"
+
+    @property
+    def description(self) -> str:
+        return "Advanced web browser automation and quality assurance testing using Playwright with VQA capabilities"
+
+    def get_available_tools(self) -> list[str]:
+        """Get the list of tools available to Web Browser Puppy."""
+        return [
+            # Core agent tools
+            "agent_share_your_reasoning",
+            # Browser control and initialization
+            "browser_initialize",
+            "browser_close",
+            "browser_status",
+            "browser_new_page",
+            "browser_list_pages",
+            # Browser navigation
+            "browser_navigate",
+            "browser_get_page_info",
+            "browser_go_back",
+            "browser_go_forward",
+            "browser_reload",
+            "browser_wait_for_load",
+            # Element discovery (semantic locators preferred)
+            "browser_find_by_role",
+            "browser_find_by_text",
+            "browser_find_by_label",
+            "browser_find_by_placeholder",
+            "browser_find_by_test_id",
+            "browser_find_buttons",
+            "browser_find_links",
+            "browser_xpath_query",  # Fallback when semantic locators fail
+            # Element interactions
+            "browser_click",
+            "browser_double_click",
+            "browser_hover",
+            "browser_set_text",
+            "browser_get_text",
+            "browser_get_value",
+            "browser_select_option",
+            "browser_check",
+            "browser_uncheck",
+            # Advanced features
+            "browser_execute_js",
+            "browser_scroll",
+            "browser_scroll_to_element",
+            "browser_set_viewport",
+            "browser_wait_for_element",
+            "browser_highlight_element",
+            "browser_clear_highlights",
+            # Screenshots and VQA
+            "browser_screenshot_analyze",
+            # Workflow management
+            "browser_save_workflow",
+            "browser_list_workflows",
+            "browser_read_workflow",
+        ]
+
+    def get_system_prompt(self) -> str:
+        """Get Web Browser Puppy's specialized system prompt."""
+        return """
+You are Quality Assurance Kitten 🐱, an advanced autonomous browser automation and QA testing agent powered by Playwright!
+
+You specialize in:
+🎯 **Quality Assurance Testing** - automated testing of web applications and user workflows
+👁️ **Visual verification** - taking screenshots and analyzing page content for bugs
+🔍 **Element discovery** - finding elements using semantic locators and accessibility best practices
+📝 **Data extraction** - scraping content and gathering information from web pages
+🧪 **Web automation** - filling forms, clicking buttons, navigating sites with precision
+🐛 **Bug detection** - identifying UI issues, broken functionality, and accessibility problems
+
+## Core Workflow Philosophy
+
+For any browser task, follow this approach:
+1. **Check Existing Workflows**: Use browser_list_workflows to see if similar tasks have been solved before
+2. **Learn from History**: If relevant workflows exist, use browser_read_workflow to review proven strategies
+3. **Plan & Reason**: Use share_your_reasoning to break down complex tasks and explain your approach
+4. **Initialize**: Always start with browser_initialize if browser isn't running
+5. **Navigate**: Use browser_navigate to reach the target page
+6. **Discover**: Use semantic locators (PREFERRED) for element discovery
+7. **Verify**: Use highlighting and screenshots to confirm elements
+8. **Act**: Interact with elements through clicks, typing, etc.
+9. **Validate**: Take screenshots or query DOM to verify actions worked
+10. **Document Success**: Use browser_save_workflow to save successful patterns for future reuse
+
+## Tool Usage Guidelines
+
+### Browser Initialization
+- **ALWAYS call browser_initialize first** before any other browser operations
+- Choose appropriate settings: headless=False for debugging, headless=True for production
+- Use browser_status to check current state
+
+### Element Discovery Best Practices (ACCESSIBILITY FIRST! 🌟)
+- **PREFER semantic locators** - they're more reliable and follow accessibility standards
+- Priority order:
+  1. browser_find_by_role (button, link, textbox, heading, etc.)
+  2. browser_find_by_label (for form inputs)
+  3. browser_find_by_text (for visible text)
+  4. browser_find_by_placeholder (for input hints)
+  5. browser_find_by_test_id (for test-friendly elements)
+  6. browser_xpath_query (ONLY as last resort)
+
+### Visual Verification Workflow
+- **Before critical actions**: Use browser_highlight_element to visually confirm
+- **After interactions**: Use browser_screenshot_analyze to verify results
+- **VQA questions**: Ask specific, actionable questions like "Is the login button highlighted?"
+
+### Form Input Best Practices
+- **ALWAYS check current values** with browser_get_value before typing
+- Use browser_get_value after typing to verify success
+- This prevents typing loops and gives clear visibility into form state
+- Clear fields when appropriate before entering new text
+
+### Error Handling & Troubleshooting
+
+**When Element Discovery Fails:**
+1. Try different semantic locators first
+2. Use browser_find_buttons or browser_find_links to see available elements
+3. Take a screenshot with browser_screenshot_analyze to understand the page layout
+4. Only use XPath as absolute last resort
+
+**When Page Interactions Fail:**
+1. Check if element is visible with browser_wait_for_element
+2. Scroll element into view with browser_scroll_to_element
+3. Use browser_highlight_element to confirm element location
+4. Try browser_execute_js for complex interactions
+
+### JavaScript Execution
+- Use browser_execute_js for:
+  - Complex page state checks
+  - Custom scrolling behavior
+  - Triggering events that standard tools can't handle
+  - Accessing browser APIs
+
+### Workflow Management 📋
+
+**ALWAYS start new tasks by checking for existing workflows!**
+
+**At the beginning of any automation task:**
+1. **browser_list_workflows** - Check what workflows are already available
+2. **browser_read_workflow** - If you find a relevant workflow, read it to understand the proven approach
+3. Adapt and apply the successful patterns from existing workflows
+
+**When to save workflows:**
+- After successfully completing a complex multi-step task
+- When you discover a reliable pattern for a common website interaction
+- After troubleshooting and finding working solutions for tricky elements
+- Include both the successful steps AND the challenges/solutions you encountered
+
+**Workflow naming conventions:**
+- Use descriptive names like "search_and_atc_walmart", "login_to_github", "fill_contact_form"
+- Include the website domain for clarity
+- Focus on the main goal/outcome
+
+**What to include in saved workflows:**
+- Step-by-step tool usage with specific parameters
+- Element discovery strategies that worked
+- Common pitfalls and how to avoid them
+- Alternative approaches for edge cases
+- Tips for handling dynamic content
+
+### Performance & Best Practices
+- Use appropriate timeouts for element discovery (default 10s is usually fine)
+- Take screenshots strategically - not after every single action
+- Use browser_wait_for_load when navigating to ensure pages are ready
+- Clear highlights when done for clean visual state
+
+## Specialized Capabilities
+
+🌐 **WCAG 2.2 Level AA Compliance**: Always prioritize accessibility in element discovery
+📸 **Visual Question Answering**: Use browser_screenshot_analyze for intelligent page analysis
+🚀 **Semantic Web Navigation**: Prefer role-based and label-based element discovery
+⚡ **Playwright Power**: Full access to modern browser automation capabilities
+📋 **Workflow Management**: Save, load, and reuse automation patterns for consistency
+
+## Important Rules
+
+- **ALWAYS check for existing workflows first** - Use browser_list_workflows at the start of new tasks
+- **ALWAYS use browser_initialize before any browser operations**
+- **PREFER semantic locators over XPath** - they're more maintainable and accessible
+- **Use visual verification for critical actions** - highlight elements and take screenshots
+- **Be explicit about your reasoning** - use share_your_reasoning for complex workflows
+- **Handle errors gracefully** - provide helpful debugging information
+- **Follow accessibility best practices** - your automation should work for everyone
+- **Document your successes** - Save working patterns with browser_save_workflow for future reuse
+
+Your browser automation should be reliable, maintainable, and accessible. You are a meticulous QA engineer who catches bugs before users do! 🐱✨
+"""

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/agents/base_agent.py

@@ -114,3 +114,12 @@ class BaseAgent(ABC):
             message_hash: Hash of a message that has been compacted/summarized.
         """
         self._compacted_message_hashes.add(message_hash)
+
+    def get_model_name(self) -> Optional[str]:
+        """Get pinned model name for this agent, if specified.
+
+        Returns:
+            Model name to use for this agent, or None to use global default.
+        """
+        from ..config import get_agent_pinned_model
+        return get_agent_pinned_model(self.name)

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/command_handler.py

@@ -81,7 +81,9 @@ def get_commands_help():
     )
     help_lines.append(
         Text("/truncate", style="cyan")
-        + Text(" <N>              Truncate message history to N most recent messages (keeping system message)")
+        + Text(
+            " <N>              Truncate message history to N most recent messages (keeping system message)"
+        )
     )
     help_lines.append(
         Text("/<unknown>", style="cyan")
@@ -409,23 +411,33 @@ def handle_command(command: str):
 
     if command.startswith("/pin_model"):
         # Handle agent model pinning
+        import json
+
         from code_puppy.agents.json_agent import discover_json_agents
         from code_puppy.command_line.model_picker_completion import load_model_names
-        import json
 
         tokens = command.split()
 
         if len(tokens) != 3:
             emit_warning("Usage: /pin_model <agent-name> <model-name>")
 
-            # Show available models and JSON agents
+            # Show available models and agents
             available_models = load_model_names()
             json_agents = discover_json_agents()
 
+            # Get built-in agents
+            from code_puppy.agents.agent_manager import get_agent_descriptions
+            builtin_agents = get_agent_descriptions()
+
             emit_info("Available models:")
             for model in available_models:
                 emit_info(f"  [cyan]{model}[/cyan]")
 
+            if builtin_agents:
+                emit_info("\nAvailable built-in agents:")
+                for agent_name, description in builtin_agents.items():
+                    emit_info(f"  [cyan]{agent_name}[/cyan] - {description}")
+
             if json_agents:
                 emit_info("\nAvailable JSON agents:")
                 for agent_name, agent_path in json_agents.items():
@@ -442,31 +454,51 @@ def handle_command(command: str):
             emit_warning(f"Available models: {', '.join(available_models)}")
             return True
 
-        # Check that we're modifying a JSON agent (not a built-in Python agent)
+        # Check if this is a JSON agent or a built-in Python agent
         json_agents = discover_json_agents()
-        if agent_name not in json_agents:
-            emit_error(f"JSON agent '{agent_name}' not found")
 
-            # Show available JSON agents
+        # Get list of available built-in agents
+        from code_puppy.agents.agent_manager import get_agent_descriptions
+        builtin_agents = get_agent_descriptions()
+
+        is_json_agent = agent_name in json_agents
+        is_builtin_agent = agent_name in builtin_agents
+
+        if not is_json_agent and not is_builtin_agent:
+            emit_error(f"Agent '{agent_name}' not found")
+
+            # Show available agents
+            if builtin_agents:
+                emit_info("Available built-in agents:")
+                for name, desc in builtin_agents.items():
+                    emit_info(f"  [cyan]{name}[/cyan] - {desc}")
+
             if json_agents:
-                emit_info("Available JSON agents:")
+                emit_info("\nAvailable JSON agents:")
                 for name, path in json_agents.items():
                     emit_info(f"  [cyan]{name}[/cyan] ({path})")
             return True
 
-        agent_file_path = json_agents[agent_name]
-
-        # Load, modify, and save the agent configuration
+        # Handle different agent types
         try:
-            with open(agent_file_path, "r", encoding="utf-8") as f:
-                agent_config = json.load(f)
+            if is_json_agent:
+                # Handle JSON agent - modify the JSON file
+                agent_file_path = json_agents[agent_name]
+
+                with open(agent_file_path, "r", encoding="utf-8") as f:
+                    agent_config = json.load(f)
 
-            # Set the model
-            agent_config["model"] = model_name
+                # Set the model
+                agent_config["model"] = model_name
 
-            # Save the updated configuration
-            with open(agent_file_path, "w", encoding="utf-8") as f:
-                json.dump(agent_config, f, indent=2, ensure_ascii=False)
+                # Save the updated configuration
+                with open(agent_file_path, "w", encoding="utf-8") as f:
+                    json.dump(agent_config, f, indent=2, ensure_ascii=False)
+
+            else:
+                # Handle built-in Python agent - store in config
+                from code_puppy.config import set_agent_pinned_model
+                set_agent_pinned_model(agent_name, model_name)
 
             emit_success(f"Model '{model_name}' pinned to agent '{agent_name}'")
 
@@ -622,9 +654,11 @@ def handle_command(command: str):
     if command.startswith("/truncate"):
         tokens = command.split()
         if len(tokens) != 2:
-            emit_error("Usage: /truncate <N> (where N is the number of messages to keep)")
+            emit_error(
+                "Usage: /truncate <N> (where N is the number of messages to keep)"
+            )
             return True
-        
+
         try:
             n = int(tokens[1])
             if n < 1:
@@ -633,23 +667,29 @@ def handle_command(command: str):
         except ValueError:
             emit_error("N must be a valid integer")
             return True
-        
+
         from code_puppy.state_management import get_message_history, set_message_history
-        
+
         history = get_message_history()
         if not history:
             emit_warning("No history to truncate yet. Ask me something first!")
             return True
-            
+
         if len(history) <= n:
-            emit_info(f"History already has {len(history)} messages, which is <= {n}. Nothing to truncate.")
+            emit_info(
+                f"History already has {len(history)} messages, which is <= {n}. Nothing to truncate."
+            )
             return True
-            
+
         # Always keep the first message (system message) and then keep the N-1 most recent messages
-        truncated_history = [history[0]] + history[-(n-1):] if n > 1 else [history[0]]
-        
+        truncated_history = (
+            [history[0]] + history[-(n - 1) :] if n > 1 else [history[0]]
+        )
+
         set_message_history(truncated_history)
-        emit_success(f"Truncated message history from {len(history)} to {len(truncated_history)} messages (keeping system message and {n-1} most recent)")
+        emit_success(
+            f"Truncated message history from {len(history)} to {len(truncated_history)} messages (keeping system message and {n - 1} most recent)"
+        )
         return True
 
     if command in ("/exit", "/quit"):

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/add_command.py

@@ -130,7 +130,7 @@ class AddCommand(MCPCommandBase):
         """
         try:
             from code_puppy.config import MCP_SERVERS_FILE
-            from code_puppy.mcp.managed_server import ServerConfig
+            from code_puppy.mcp_.managed_server import ServerConfig
 
             # Extract required fields
             name = config_dict.pop("name")

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/base.py

@@ -8,7 +8,7 @@ import logging
 
 from rich.console import Console
 
-from code_puppy.mcp.manager import get_mcp_manager
+from code_puppy.mcp_.manager import get_mcp_manager
 
 # Configure logging
 logger = logging.getLogger(__name__)

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/install_command.py

@@ -76,7 +76,7 @@ class InstallCommand(MCPCommandBase):
     def _install_from_catalog(self, server_name_or_id: str, group_id: str) -> bool:
         """Install a server directly from the catalog by name or ID."""
         try:
-            from code_puppy.mcp.server_registry_catalog import catalog
+            from code_puppy.mcp_.server_registry_catalog import catalog
             from code_puppy.messaging import emit_prompt
 
             from .utils import find_server_id_by_name

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/list_command.py

@@ -8,7 +8,7 @@ from typing import List, Optional
 from rich.table import Table
 from rich.text import Text
 
-from code_puppy.mcp.managed_server import ServerState
+from code_puppy.mcp_.managed_server import ServerState
 from code_puppy.messaging import emit_info
 
 from .base import MCPCommandBase

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/search_command.py

@@ -34,7 +34,7 @@ class SearchCommand(MCPCommandBase):
             group_id = self.generate_group_id()
 
         try:
-            from code_puppy.mcp.server_registry_catalog import catalog
+            from code_puppy.mcp_.server_registry_catalog import catalog
 
             if not args:
                 # Show popular servers if no query

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/start_all_command.py

@@ -6,7 +6,7 @@ import logging
 import time
 from typing import List, Optional
 
-from code_puppy.mcp.managed_server import ServerState
+from code_puppy.mcp_.managed_server import ServerState
 from code_puppy.messaging import emit_info
 
 from .base import MCPCommandBase

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/status_command.py

@@ -8,7 +8,7 @@ from typing import List, Optional
 
 from rich.panel import Panel
 
-from code_puppy.mcp.managed_server import ServerState
+from code_puppy.mcp_.managed_server import ServerState
 from code_puppy.messaging import emit_info
 
 from .base import MCPCommandBase
@@ -117,7 +117,7 @@ class StatusCommand(MCPCommandBase):
 
             # Check async lifecycle manager status if available
             try:
-                from code_puppy.mcp.async_lifecycle import get_lifecycle_manager
+                from code_puppy.mcp_.async_lifecycle import get_lifecycle_manager
 
                 lifecycle_mgr = get_lifecycle_manager()
                 if lifecycle_mgr.is_running(server_id):

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/stop_all_command.py

@@ -6,7 +6,7 @@ import logging
 import time
 from typing import List, Optional
 
-from code_puppy.mcp.managed_server import ServerState
+from code_puppy.mcp_.managed_server import ServerState
 from code_puppy.messaging import emit_info
 
 from .base import MCPCommandBase

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/utils.py

@@ -8,7 +8,7 @@ from typing import Optional
 
 from rich.text import Text
 
-from code_puppy.mcp.managed_server import ServerState
+from code_puppy.mcp_.managed_server import ServerState
 
 
 def format_state_indicator(state: ServerState) -> Text:

{code_puppy-0.0.171 → code_puppy-0.0.172}/code_puppy/command_line/mcp/wizard_utils.py

@@ -118,7 +118,7 @@ def interactive_server_selection(group_id: str):
     # This is a simplified version - the full implementation would have
     # category browsing, search, etc. For now, we'll just show popular servers
     try:
-        from code_puppy.mcp.server_registry_catalog import catalog
+        from code_puppy.mcp_.server_registry_catalog import catalog
 
         servers = catalog.get_popular(10)
         if not servers:
@@ -256,7 +256,7 @@ def install_server_from_catalog(
         import os
 
         from code_puppy.config import MCP_SERVERS_FILE
-        from code_puppy.mcp.managed_server import ServerConfig
+        from code_puppy.mcp_.managed_server import ServerConfig
 
         # Set environment variables in the current environment
         for var, value in env_vars.items():