hud-python 0.4.57__py3-none-any.whl → 0.4.59__py3-none-any.whl

hud/cli/dev.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import asyncio
+import contextlib
 import importlib
 import importlib.util
 import logging
@@ -13,6 +14,8 @@ import threading
 from pathlib import Path
 from typing import Any
 
+import typer
+
 from hud.utils.hud_console import HUDConsole
 
 hud_console = HUDConsole()
@@ -26,6 +29,7 @@ def show_dev_server_info(
     interactive: bool,
     env_dir: Path | None = None,
     new: bool = False,
+    docker_mode: bool = False,
 ) -> str:
     """Show consistent server info for both Python and Docker modes.
 
@@ -54,7 +58,15 @@ def show_dev_server_info(
     if transport == "http":
         hud_console.section_title("Quick Links")
         hud_console.info(f"{hud_console.sym.ITEM} Docs: http://localhost:{port}/docs")
-        hud_console.info(f"{hud_console.sym.ITEM} Cursor: {cursor_deeplink}")
+        hud_console.info(f"{hud_console.sym.ITEM} Cursor:")
+        # Display the Cursor link on its own line to prevent wrapping
+        hud_console.link(cursor_deeplink)
+
+        # Show eval endpoint if in Docker mode
+        if docker_mode:
+            hud_console.info(
+                f"{hud_console.sym.ITEM} Eval API: http://localhost:{port}/eval (POST)"
+            )
 
         # Check for VNC (browser environment)
         if env_dir and (env_dir / "environment" / "server.py").exists():
@@ -510,6 +522,9 @@ def run_docker_dev_server(
     new: bool = False,
 ) -> None:
     """Run MCP server in Docker with volume mounts, expose via local HTTP proxy."""
+    import atexit
+    import signal
+
     import typer
     import yaml
 
@@ -522,6 +537,69 @@ def run_docker_dev_server(
 
     cwd = Path.cwd()
 
+    # Container name will be set later and used for cleanup
+    container_name: str | None = None
+    cleanup_done = False
+
+    def cleanup_container() -> None:
+        """Clean up Docker container on exit."""
+        nonlocal cleanup_done
+        if cleanup_done or not container_name:
+            return
+
+        cleanup_done = True
+        hud_console.debug(f"Cleaning up container: {container_name}")
+
+        # Check if container is still running
+        try:
+            result = subprocess.run(  # noqa: S603
+                ["docker", "ps", "-q", "-f", f"name={container_name}"],  # noqa: S607
+                stdout=subprocess.PIPE,
+                stderr=subprocess.DEVNULL,
+                text=True,
+                timeout=5,
+            )
+            if not result.stdout.strip():
+                # Container is not running, just try to remove it
+                subprocess.run(  # noqa: S603
+                    ["docker", "rm", "-f", container_name],  # noqa: S607
+                    stdout=subprocess.DEVNULL,
+                    stderr=subprocess.DEVNULL,
+                    timeout=5,
+                )
+                return
+        except Exception:  # noqa: S110
+            pass
+
+        try:
+            # First try to stop gracefully
+            subprocess.run(  # noqa: S603
+                ["docker", "stop", container_name],  # noqa: S607
+                stdout=subprocess.DEVNULL,
+                stderr=subprocess.DEVNULL,
+                timeout=10,
+            )
+            hud_console.debug(f"Container {container_name} stopped successfully")
+        except subprocess.TimeoutExpired:
+            # Force kill if stop times out
+            hud_console.debug(f"Container {container_name} stop timeout, forcing kill")
+            with contextlib.suppress(Exception):
+                subprocess.run(  # noqa: S603
+                    ["docker", "kill", container_name],  # noqa: S607
+                    stdout=subprocess.DEVNULL,
+                    stderr=subprocess.DEVNULL,
+                    timeout=5,
+                )
+
+    # Set up signal handlers for cleanup
+    def signal_handler(signum: int, frame: Any) -> None:
+        cleanup_container()
+        sys.exit(0)
+
+    signal.signal(signal.SIGTERM, signal_handler)
+    if sys.platform != "win32":
+        signal.signal(signal.SIGHUP, signal_handler)
+
     # Find environment directory (current or parent with hud.lock.yaml)
     env_dir = cwd
     lock_path = env_dir / "hud.lock.yaml"
@@ -562,10 +640,14 @@ def run_docker_dev_server(
     base_name = image_name.replace(":", "-").replace("/", "-")
     container_name = f"{base_name}-dev-{pid}"
 
+    # Register cleanup function with atexit
+    atexit.register(cleanup_container)
+
     # Build docker run command with volume mounts and folder-mode envs
     from .utils.docker import create_docker_run_command
 
     base_args = [
+        "--rm",  # Automatically remove container when it stops
         "--name",
         container_name,
         "-v",
@@ -643,6 +725,7 @@ def run_docker_dev_server(
             interactive=interactive,
             env_dir=env_dir,
             new=new,
+            docker_mode=True,
         )
         hud_console.dim_info(
             "",
@@ -679,6 +762,11 @@ def run_docker_dev_server(
             os.environ["_HUD_DEV_DOCKER_CONTAINER"] = container_name
             hud_console.debug(f"Docker container: {container_name}")
 
+        # Store the docker mcp_config for the eval endpoint
+        import json
+
+        os.environ["_HUD_DEV_DOCKER_MCP_CONFIG"] = json.dumps(mcp_config)
+
         # Create FastMCP proxy using the ProxyClient
         fastmcp_proxy = FastMCP.as_proxy(proxy_client)
 
@@ -713,7 +801,15 @@ def run_docker_dev_server(
         asyncio.run(run_proxy())
     except KeyboardInterrupt:
         hud_console.info("\n\nStopping...")
+        cleanup_container()
         raise typer.Exit(0) from None
+    except Exception:
+        # Ensure cleanup happens on any exception
+        cleanup_container()
+        raise
+    finally:
+        # Final cleanup attempt
+        cleanup_container()
 
 
 def run_mcp_dev_server(
@@ -732,6 +828,20 @@ def run_mcp_dev_server(
     docker_args = docker_args or []
     cwd = Path.cwd()
 
+    # Find an available port if not using stdio transport
+    if not stdio:
+        from hud.cli.utils.logging import find_free_port
+
+        actual_port = find_free_port(port)
+        if actual_port is None:
+            hud_console.error(f"No available ports found starting from {port}")
+            raise typer.Exit(1)
+
+        if actual_port != port:
+            hud_console.info(f"Port {port} is in use, using port {actual_port} instead")
+
+        port = actual_port
+
     # Auto-detect Docker mode if Dockerfile present and no module specified
     if not docker and module is None and should_use_docker_mode(cwd):
         hud_console.note("Detected Dockerfile - using Docker mode with volume mounts")

hud/cli/eval.py

@@ -53,7 +53,7 @@ def get_available_models() -> list[dict[str, str | None]]:
     try:
         from hud.cli.rl import rl_api
 
-        hud_console.info("Fetching your models from https://hud.so/models")
+        hud_console.info("Fetching your models from https://hud.ai/models")
         models = rl_api.list_models()
 
         # Filter for ready models only and sort by recency
@@ -188,6 +188,24 @@ def build_agent(
         else:
             return OperatorAgent(verbose=verbose)
 
+    elif agent_type == AgentType.GEMINI:
+        try:
+            from hud.agents import GeminiAgent
+        except ImportError as e:
+            hud_console.error(
+                "Gemini agent dependencies are not installed. "
+                "Please install with: pip install 'hud-python[agent]'"
+            )
+            raise typer.Exit(1) from e
+
+        gemini_kwargs: dict[str, Any] = {
+            "model": model or "gemini-2.5-computer-use-preview-10-2025",
+            "verbose": verbose,
+        }
+        if allowed_tools:
+            gemini_kwargs["allowed_tools"] = allowed_tools
+        return GeminiAgent(**gemini_kwargs)
+
     elif agent_type == AgentType.LITELLM:
         try:
             from hud.agents.lite_llm import LiteAgent
@@ -344,6 +362,17 @@ async def run_single_task(
         agent_config = {"verbose": verbose}
         if allowed_tools:
             agent_config["allowed_tools"] = allowed_tools
+    elif agent_type == AgentType.GEMINI:
+        from hud.agents import GeminiAgent
+
+        agent_class = GeminiAgent
+        agent_config = {
+            "model": model or "gemini-2.5-computer-use-preview-10-2025",
+            "verbose": verbose,
+            "validate_api_key": False,
+        }
+        if allowed_tools:
+            agent_config["allowed_tools"] = allowed_tools
     elif agent_type == AgentType.LITELLM:
         from hud.agents.lite_llm import LiteAgent
 
@@ -534,6 +563,26 @@ async def run_full_dataset(
         if allowed_tools:
             agent_config["allowed_tools"] = allowed_tools
 
+    elif agent_type == AgentType.GEMINI:
+        try:
+            from hud.agents import GeminiAgent
+
+            agent_class = GeminiAgent
+        except ImportError as e:
+            hud_console.error(
+                "Gemini agent dependencies are not installed. "
+                "Please install with: pip install 'hud-python[agent]'"
+            )
+            raise typer.Exit(1) from e
+
+        agent_config = {
+            "model": model or "gemini-2.5-computer-use-preview-10-2025",
+            "verbose": verbose,
+            "validate_api_key": False,
+        }
+        if allowed_tools:
+            agent_config["allowed_tools"] = allowed_tools
+
     elif agent_type == AgentType.LITELLM:
         try:
             from hud.agents.lite_llm import LiteAgent
@@ -641,7 +690,7 @@ def eval_command(
     agent: AgentType = typer.Option(  # noqa: B008
         AgentType.CLAUDE,
         "--agent",
-        help="Agent backend to use (claude, openai, vllm for local server, or litellm)",
+        help="Agent backend to use (claude, gemini, openai, vllm for local servers, or litellm)",
     ),
     model: str | None = typer.Option(
         None,
@@ -757,6 +806,13 @@ def eval_command(
                 "Set it in your environment or run: hud set ANTHROPIC_API_KEY=your-key-here"
             )
             raise typer.Exit(1)
+    elif agent == AgentType.GEMINI:
+        if not settings.gemini_api_key:
+            hud_console.error("GEMINI_API_KEY is required for Gemini agent")
+            hud_console.info(
+                "Set it in your environment or run: hud set GEMINI_API_KEY=your-key-here"
+            )
+            raise typer.Exit(1)
     elif agent == AgentType.OPENAI and not settings.openai_api_key:
         hud_console.error("OPENAI_API_KEY is required for OpenAI agent")
         hud_console.info("Set it in your environment or run: hud set OPENAI_API_KEY=your-key-here")
@@ -771,7 +827,7 @@ def eval_command(
     # Check for HUD_API_KEY if using HUD services
     if not settings.api_key:
         hud_console.warning("HUD_API_KEY not set. Some features may be limited.")
-        hud_console.info("Get your API key at: https://hud.so")
+        hud_console.info("Get your API key at: https://hud.ai")
         hud_console.info("Set it in your environment or run: hud set HUD_API_KEY=your-key-here")
 
     # Parse allowed tools

hud/cli/flows/dev.py

@@ -47,11 +47,11 @@ async def create_dynamic_trace(
 
     try:
         resp = await make_request("POST", url=url, json=payload, api_key=api_key)
-        # New API returns an id; construct the URL as https://hud.so/trace/{id}
+        # New API returns an id; construct the URL as https://hud.ai/trace/{id}
         trace_id = resp.get("id")
 
         if isinstance(trace_id, str) and trace_id:
-            return trace_id, f"https://hud.so/trace/{trace_id}"
+            return trace_id, f"https://hud.ai/trace/{trace_id}"
         return None, None
     except Exception as e:
         # Do not interrupt dev flow
@@ -114,7 +114,9 @@ def show_dev_ui(
     label = "Base image" if is_docker else "Server"
     hud_console.info("")
     hud_console.info(f"{hud_console.sym.ITEM} {label}: {server_name}")
-    hud_console.info(f"{hud_console.sym.ITEM} Cursor: {cursor_deeplink}")
+    hud_console.info(f"{hud_console.sym.ITEM} Cursor:")
+    # Display the Cursor link on its own line to prevent wrapping
+    hud_console.link(cursor_deeplink)
     hud_console.info("")
     hud_console.info(f"{hud_console.sym.SUCCESS} Hot-reload enabled")
     if is_docker:

hud/cli/init.py

@@ -182,17 +182,17 @@ def create_environment(
 
     hud_console = HUDConsole()
 
-    # Determine environment name/target directory
-    if name is None:
-        current_dir = Path.cwd()
-        name = current_dir.name
-        target_dir = current_dir
-        hud_console.info(f"Using current directory name: {name}")
-    else:
-        target_dir = Path(directory) / name
-
     # Choose preset
     preset_normalized = (preset or "").strip().lower() if preset else _prompt_for_preset()
+
+    # If no name is provided, use the preset name as the environment name
+    if name is None:
+        name = preset_normalized
+        hud_console.info(f"Using preset name as environment name: {name}")
+
+    # Always create a new directory based on the name
+    target_dir = Path.cwd() / name if directory == "." else Path(directory) / name
+
     if preset_normalized not in PRESET_MAP:
         hud_console.warning(
             f"Unknown preset '{preset_normalized}', defaulting to 'blank' "
@@ -263,14 +263,10 @@ def create_environment(
         hud_console.status_item(entry, "added")
 
     hud_console.section_title("Next steps")
-    if target_dir == Path.cwd():
-        hud_console.info("1. Start development server (with MCP inspector):")
-        hud_console.command_example("hud dev --inspector")
-    else:
-        hud_console.info("1. Enter the directory:")
-        hud_console.command_example(f"cd {target_dir}")
-        hud_console.info("\n2. Start development server (with MCP inspector):")
-        hud_console.command_example("hud dev --inspector")
-
+    # Since we now almost always create a new directory, show cd command
+    hud_console.info("1. Enter the directory:")
+    hud_console.command_example(f"cd {target_dir.name}")
+    hud_console.info("\n2. Start development server (with MCP inspector):")
+    hud_console.command_example("hud dev --inspector")
     hud_console.info("\n3. Review the README in this preset for specific instructions.")
     hud_console.info("\n4. Customize as needed.")

hud/cli/push.py

@@ -152,7 +152,7 @@ def push_environment(
         hud_console.error("No HUD API key found")
         hud_console.warning("A HUD API key is required to push environments.")
         hud_console.info("\nTo get started:")
-        hud_console.info("1. Get your API key at: https://hud.so/settings")
+        hud_console.info("1. Get your API key at: https://hud.ai/settings")
         hud_console.info("Set it in your environment or run: hud set HUD_API_KEY=your-key-here")
         hud_console.command_example("hud push", "Try again")
         hud_console.info("")
@@ -440,7 +440,7 @@ def push_environment(
         elif response.status_code == 401:
             hud_console.error("Authentication failed")
             hud_console.info("Check your HUD_API_KEY is valid")
-            hud_console.info("Get a new key at: https://hud.so/settings")
+            hud_console.info("Get a new key at: https://hud.ai/settings")
             hud_console.info("Set it in your environment or run: hud set HUD_API_KEY=your-key-here")
         elif response.status_code == 403:
             hud_console.error("Permission denied")

hud/cli/rl/__init__.py

@@ -25,7 +25,7 @@ def rl_command(
     ),
     model: str | None = typer.Argument(
         None,
-        help="Model to train from https://hud.so/models (default: interactive selection)",
+        help="Model to train from https://hud.ai/models (default: interactive selection)",
     ),
     config_file: Path | None = typer.Option(  # noqa: B008
         None,

hud/cli/rl/celebrate.py

@@ -133,7 +133,7 @@ def show_confetti(console: Console, seconds: float = 2.5) -> None:
     """
     # Show celebratory message first
     console.print(
-        "[bold green]🎉 Starting training! See your model on https://hud.so/models[/bold green]"
+        "[bold green]🎉 Starting training! See your model on https://hud.ai/models[/bold green]"
     )
     time.sleep(0.3)  # Brief pause to see the message
 

hud/cli/rl/remote_runner.py

@@ -55,7 +55,7 @@ def ensure_vllm_deployed(
     hud_console.info("Waiting for vLLM server to be ready...")
     start_time = time.time()
     with hud_console.progress() as progress:
-        progress.update("Checking deployment status (see live status on https://hud.so/models)")
+        progress.update("Checking deployment status (see live status on https://hud.ai/models)")
         while True:
             if time.time() - start_time > timeout:
                 hud_console.error("Timeout waiting for vLLM deployment")
@@ -139,7 +139,7 @@ def run_remote_training(
     hud_console.section_title("Model Selection")
 
     # Fetch existing models
-    hud_console.info("Fetching your models from https://hud.so/models")
+    hud_console.info("Fetching your models from https://hud.ai/models")
 
     try:
         models = rl_api.list_models()
@@ -312,7 +312,7 @@ def run_remote_training(
         # gpu_table.add_column("Price/hr", style="yellow")
 
         # for gpu, info in GPU_PRICING.items():
-        #     gpu_table.add_row(gpu, info["memory"], "see pricing on hud.so")
+        #     gpu_table.add_row(gpu, info["memory"], "see pricing on hud.ai")
 
         # console.print(gpu_table)
 

hud/cli/tests/test_eval.py

@@ -68,6 +68,26 @@ class TestBuildAgent:
             )
             assert result == mock_instance
 
+    def test_builds_gemini_agent(self) -> None:
+        """Test building a Gemini agent."""
+        with patch("hud.agents.GeminiAgent") as mock_runner:
+            mock_instance = Mock()
+            mock_runner.return_value = mock_instance
+
+            result = build_agent(
+                AgentType.GEMINI,
+                model="gemini-test",
+                allowed_tools=["gemini_computer"],
+                verbose=True,
+            )
+
+            mock_runner.assert_called_once_with(
+                model="gemini-test",
+                verbose=True,
+                allowed_tools=["gemini_computer"],
+            )
+            assert result == mock_instance
+
 
 class TestRunSingleTask:
     """Test the run_single_task function."""

hud/clients/base.py

@@ -140,7 +140,7 @@ class BaseHUDClient(AgentMCPClient):
                     raise HudAuthenticationError(
                         f'Sending authorization "{headers.get("Authorization", "")}", which may'
                         " be incomplete. Ensure HUD_API_KEY environment variable is set or send it"
-                        " as a header. You can get an API key at https://hud.so"
+                        " as a header. You can get an API key at https://hud.ai"
                     )
             # Subclasses implement connection
             await self._connect(self._mcp_config)

hud/clients/fastmcp.py

@@ -95,7 +95,7 @@ class FastMCPHUDClient(BaseHUDClient):
                             raise RuntimeError(
                                 "Authentication failed for HUD API. "
                                 "Please ensure your HUD_API_KEY environment variable is set correctly."  # noqa: E501
-                                "You can get an API key at https://hud.so"
+                                "You can get an API key at https://hud.ai"
                             ) from e
                     # Generic 401 error
                     raise RuntimeError(

hud/otel/config.py

@@ -113,7 +113,7 @@ def configure_telemetry(
         # Error if no exporters are configured
         raise ValueError(
             "No telemetry backend configured. Either:\n"
-            "1. Set HUD_API_KEY environment variable for HUD telemetry (https://hud.so)\n"
+            "1. Set HUD_API_KEY environment variable for HUD telemetry (https://hud.ai)\n"
             "2. Use enable_otlp=True with configure_telemetry() for alternative backends (e.g., Jaeger)\n"  # noqa: E501
         )
     elif not settings.telemetry_enabled:

hud/otel/context.py

@@ -408,7 +408,7 @@ def _print_trace_url(task_run_id: str) -> None:
     if not (settings.telemetry_enabled and settings.api_key):
         return
 
-    url = f"https://hud.so/trace/{task_run_id}"
+    url = f"https://hud.ai/trace/{task_run_id}"
     header = "🚀 See your agent live at:"
 
     # ANSI color codes
@@ -447,7 +447,7 @@ def _print_trace_complete_url(task_run_id: str, error_occurred: bool = False) ->
     if not (settings.telemetry_enabled and settings.api_key):
         return
 
-    url = f"https://hud.so/trace/{task_run_id}"
+    url = f"https://hud.ai/trace/{task_run_id}"
 
     # ANSI color codes
     GREEN = "\033[92m"