hud-python 0.1.0b2__py3-none-any.whl → 0.1.1__py3-none-any.whl

hud/__init__.py

@@ -9,7 +9,7 @@ from hud.environment import Environment, EvalSet, Observation, TaskResult
 from hud.gym import Gym
 from hud.run import Run
 
-__version__ = "0.1.0b2"
+__version__ = "0.1.1"
 
 __all__ = [
     "Environment",

hud/adapters/claude/__init__.py

@@ -3,4 +3,3 @@ from __future__ import annotations
 from .adapter import ClaudeAdapter
 
 __all__ = ["ClaudeAdapter"]
-

hud/adapters/claude/adapter.py

@@ -2,10 +2,11 @@
 
 from __future__ import annotations
 
-from typing import Any
+from typing import Any, ClassVar
 
 from hud.adapters.common import CLA, Adapter
 from hud.adapters.common.types import (
+    CLAKey,
     ClickAction,
     DragAction,
     MoveAction,
@@ -20,11 +21,17 @@ from hud.adapters.common.types import (
 
 
 class ClaudeAdapter(Adapter):
+    KEY_MAP: ClassVar[dict[str, CLAKey]] = {"Return": "enter"}
+
     def __init__(self) -> None:
         super().__init__()
         self.agent_width = 1024  # Claude's preferred width
         self.agent_height = 768  # Claude's preferred height
 
+    def _map_key(self, key: str) -> CLAKey:
+        """Map a key to its standardized form."""
+        return self.KEY_MAP.get(key, key.lower())  # type: ignore
+
     def convert(self, data: Any) -> CLA:
         try:
             action_type = data.get("action")
@@ -32,10 +39,12 @@ class ClaudeAdapter(Adapter):
             if action_type == "key":
                 assert "text" in data
                 if "+" in data["text"]:
-                    keys = data["text"].split("+")
+                    keys: list[CLAKey] = [
+                        self._map_key(k) for k in (data["text"].split("+"))
+                    ]
                     assert len(keys) > 0
                     return PressAction(keys=keys)
-                return PressAction(keys=[data["text"]])
+                return PressAction(keys=[self._map_key(data["text"])])
 
             elif action_type == "type":
                 assert "text" in data
@@ -66,12 +75,19 @@ class ClaudeAdapter(Adapter):
                 assert len(coord) == 2
                 if (
                     len(self.memory) == 0
-                    or (self.memory[-1] is not MoveAction and self.memory[-1] is not ClickAction)
+                    or (
+                        self.memory[-1] is not MoveAction
+                        and self.memory[-1] is not ClickAction
+                    )
                     or self.memory[-1].point is None
                 ):
-                    raise ValueError("Left click drag must be preceded by a move or click action")
+                    raise ValueError(
+                        "Left click drag must be preceded by a move or click action"
+                    )
                 else:
-                    return DragAction(path=[self.memory[-1].point, Point(x=coord[0], y=coord[1])])
+                    return DragAction(
+                        path=[self.memory[-1].point, Point(x=coord[0], y=coord[1])]
+                    )
 
             elif action_type == "right_click":
                 assert "coordinate" in data
@@ -96,6 +112,17 @@ class ClaudeAdapter(Adapter):
                     point=Point(x=coord[0], y=coord[1]), button="left", pattern=[100]
                 )
 
+            elif action_type == "triple_click":
+                assert "coordinate" in data
+                coord = data["coordinate"]
+                assert isinstance(coord, list)
+                assert len(coord) == 2
+                return ClickAction(
+                    point=Point(x=coord[0], y=coord[1]),
+                    button="left",
+                    pattern=[100, 100],
+                )
+
             elif action_type == "scroll":
                 assert "scroll_direction" in data
                 direction = data["scroll_direction"]
@@ -112,7 +139,8 @@ class ClaudeAdapter(Adapter):
                     raise ValueError(f"Unsupported scroll direction: {direction}")
 
                 return ScrollAction(
-                    point=Point(x=data["coordinate"][0], y=data["coordinate"][1]), scroll=scroll
+                    point=Point(x=data["coordinate"][0], y=data["coordinate"][1]),
+                    scroll=scroll,
                 )
 
             elif action_type == "screenshot":
@@ -124,7 +152,6 @@ class ClaudeAdapter(Adapter):
             elif action_type == "wait":
                 assert "duration" in data
                 return WaitAction(time=data["duration"])
-
             else:
                 raise ValueError(f"Unsupported action type: {action_type}")
         except AssertionError:

hud/adapters/common/types.py

@@ -23,12 +23,13 @@ class ClickAction(CLAAction):
     selector: str | None = None
     button: Literal["left", "right", "wheel", "back", "forward"] = "left"
     pattern: list[int] | None = None  # [delay_1, delay_2, ...]
+    hold_keys: list[CLAKey] | None = None
 
 
 # PRESS ACTION for key presses/hotkeys
 class PressAction(CLAAction):
     type: Literal["press"] = "press"
-    keys: list[str]
+    keys: list[CLAKey]
 
 
 # TYPE ACTION for text typing
@@ -43,6 +44,7 @@ class ScrollAction(CLAAction):
     type: Literal["scroll"] = "scroll"
     point: Point | None = None
     scroll: Point | None = None
+    hold_keys: list[CLAKey] | None = None
 
 
 # MOVE ACTION for mouse movement
@@ -64,6 +66,7 @@ class DragAction(CLAAction):
     type: Literal["drag"] = "drag"
     path: list[Point]
     pattern: list[int] | None = None  # [delay_1, delay_2, ...]
+    hold_keys: list[CLAKey] | None = None
 
 
 # SCREENSHOT ACTION
@@ -90,3 +93,201 @@ CLA = Annotated[
     ],
     Field(discriminator="type"),
 ]
+
+
+CLAKey = Literal[
+    # Control keys
+    "backspace",
+    "tab",
+    "enter",
+    "shift",
+    "shiftleft",
+    "shiftright",
+    "ctrl",
+    "ctrlleft",
+    "ctrlright",
+    "alt",
+    "altleft",
+    "altright",
+    "pause",
+    "capslock",
+    "esc",
+    "escape",
+    "space",
+    "pageup",
+    "pagedown",
+    "end",
+    "home",
+    "left",
+    "up",
+    "right",
+    "down",
+    "select",
+    "print",
+    "execute",
+    "printscreen",
+    "prtsc",
+    "insert",
+    "delete",
+    "help",
+    "sleep",
+    # Special keys
+    "numlock",
+    "scrolllock",
+    "clear",
+    "separator",
+    "modechange",
+    "apps",
+    "browserback",
+    "browserfavorites",
+    "browserforward",
+    "browserhome",
+    "browserrefresh",
+    "browsersearch",
+    "browserstop",
+    "launchapp1",
+    "launchapp2",
+    "launchmail",
+    "launchmediaselect",
+    "playpause",
+    "stop",
+    "prevtrack",
+    "nexttrack",
+    "volumemute",
+    "volumeup",
+    "volumedown",
+    "zoom",
+    # Modifier keys
+    "win",
+    "winleft",
+    "winright",
+    "command",
+    "option",
+    "optionleft",
+    "optionright",
+    "fn",
+    # Numpad keys
+    "num0",
+    "num1",
+    "num2",
+    "num3",
+    "num4",
+    "num5",
+    "num6",
+    "num7",
+    "num8",
+    "num9",
+    "multiply",
+    "add",
+    "subtract",
+    "decimal",
+    "divide",
+    # Function keys
+    "f1",
+    "f2",
+    "f3",
+    "f4",
+    "f5",
+    "f6",
+    "f7",
+    "f8",
+    "f9",
+    "f10",
+    "f11",
+    "f12",
+    "f13",
+    "f14",
+    "f15",
+    "f16",
+    "f17",
+    "f18",
+    "f19",
+    "f20",
+    "f21",
+    "f22",
+    "f23",
+    "f24",
+    # Language-specific keys
+    "hanguel",
+    "hangul",
+    "hanja",
+    "kana",
+    "kanji",
+    "junja",
+    "convert",
+    "nonconvert",
+    "yen",
+    # Characters
+    "\t",
+    "\n",
+    "\r",
+    " ",
+    "!",
+    '"',
+    "#",
+    "$",
+    "%",
+    "&",
+    "'",
+    "(",
+    ")",
+    "*",
+    "+",
+    ",",
+    "-",
+    ".",
+    "/",
+    "0",
+    "1",
+    "2",
+    "3",
+    "4",
+    "5",
+    "6",
+    "7",
+    "8",
+    "9",
+    ":",
+    ";",
+    "<",
+    "=",
+    ">",
+    "?",
+    "@",
+    "[",
+    "\\",
+    "]",
+    "^",
+    "_",
+    "`",
+    "a",
+    "b",
+    "c",
+    "d",
+    "e",
+    "f",
+    "g",
+    "h",
+    "i",
+    "j",
+    "k",
+    "l",
+    "m",
+    "n",
+    "o",
+    "p",
+    "q",
+    "r",
+    "s",
+    "t",
+    "u",
+    "v",
+    "w",
+    "x",
+    "y",
+    "z",
+    "{",
+    "|",
+    "}",
+    "~",
+]

hud/client.py

@@ -11,14 +11,14 @@ from .adapters.common import Adapter
 from .environment import EvalSet
 from .gym import Gym
 from .run import Run, RunResponse
-from .server import make_request, make_sync_request
+from .server import make_request
 from .settings import settings
 
 
 class HUDClient:
     """
     Client for interacting with the HUD API.
-    
+
     This is the main entry point for the SDK, providing methods to load gyms,
     evalsets, and create runs.
     """
@@ -26,7 +26,7 @@ class HUDClient:
     def __init__(self, api_key: str | None = None) -> None:
         """
         Initialize the HUD client with an API key.
-        
+
         Args:
             api_key: API key for authentication with the HUD API
         """
@@ -36,10 +36,10 @@ class HUDClient:
     async def load_gym(self, id: str) -> Gym:
         """
         Load a gym by ID from the HUD API.
-        
+
         Args:
             id: The ID of the gym to load
-            
+
         Returns:
             Gym: The loaded gym object
         """
@@ -54,10 +54,10 @@ class HUDClient:
     async def load_evalset(self, id: str) -> EvalSet:
         """
         Load an evalset by ID from the HUD API.
-        
+
         Args:
             id: The ID of the evalset to load
-            
+
         Returns:
             EvalSet: The loaded evalset object
         """
@@ -72,7 +72,7 @@ class HUDClient:
     async def list_gyms(self) -> list[str]:
         """
         List all available gyms.
-        
+
         Returns:
             list[str]: List of gym IDs
         """
@@ -85,7 +85,7 @@ class HUDClient:
     async def get_runs(self) -> list[Run]:
         """
         Get all runs associated with the API key.
-        
+
         Returns:
             list[Run]: List of run objects
         """
@@ -98,11 +98,11 @@ class HUDClient:
     async def load_run(self, id: str, adapter: Adapter | None = None) -> Run | None:
         """
         Load a run by ID from the HUD API.
-        
+
         Args:
             id: The ID of the run to load
             adapter: Optional adapter for action conversion
-            
+
         Returns:
             Run: The loaded run object, or None if not found
         """
@@ -132,7 +132,7 @@ class HUDClient:
             )
         return None
 
-    def create_run(
+    async def create_run(
         self,
         name: str,
         gym: Gym,
@@ -143,7 +143,7 @@ class HUDClient:
     ) -> Run:
         """
         Create a new run in the HUD system.
-        
+
         Args:
             name: Name of the run
             gym: Gym to use for the run
@@ -151,7 +151,7 @@ class HUDClient:
             config: Optional configuration parameters
             metadata: Optional metadata for the run
             adapter: Optional adapter for action conversion
-            
+
         Returns:
             Run: The created run object
         """
@@ -161,7 +161,7 @@ class HUDClient:
             metadata = {}
         if config is None:
             config = {}
-        data = make_sync_request(
+        data = await make_request(
             method="POST",
             url=f"{settings.base_url}/runs",
             json={
@@ -188,6 +188,7 @@ class HUDClient:
         Display a stream in the HUD system.
         """
         from IPython.display import HTML, display
+
         html_content = f"""
         <div style="width: 960px; height: 540px; overflow: hidden;">
             <div style="transform: scale(0.5); transform-origin: top left;">

hud/environment.py

@@ -13,16 +13,29 @@ from hud.settings import settings
 if TYPE_CHECKING:
     from .adapters.common import Adapter
 
+
+class BaseResponseWithLogs(BaseModel):
+    """Base model for API responses that include logs."""
+    logs: str | None = None
+    error: str | None = None
+
+
+class RewardResponse(BaseResponseWithLogs):
+    reward: float
+
+
 logger = logging.getLogger("hud.environment")
 
+
 class Observation(BaseModel):
     """
     Observation from the environment.
-    
+
     Attributes:
         screenshot: Base64 encoded PNG string of the screen
         text: Text observation, if available
     """
+
     screenshot: str | None = None  # base64 string png
     text: str | None = None
 
@@ -30,18 +43,20 @@ class Observation(BaseModel):
 class TaskResult(BaseModel):
     """
     Result of a task step.
-    
+
     Attributes:
         observation: The current observation
         reward: Reward value from the step
         terminated: Whether the task is complete
         info: Additional information from the environment
     """
+
     observation: Observation
     reward: float
     terminated: bool
     info: dict[str, Any]
 
+
 class EnvironmentStatus(str, enum.Enum):
     """
     Status of the environment.
@@ -52,6 +67,7 @@ class EnvironmentStatus(str, enum.Enum):
         COMPLETED: The environment is completed
         ERROR: The environment is in an error state
     """
+
     INITIALIZING = "initializing"
     RUNNING = "running"
     COMPLETED = "completed"
@@ -64,27 +80,30 @@ status_messages = {
     EnvironmentStatus.COMPLETED.value: "completed",
 }
 
+
 class Environment:
     """
     Environment interface for agent interactions.
-    
+
     This class handles the environment state and interactions, including
     creating the environment, retrieving state, and executing actions.
     """
 
     def __init__(
         self,
-        run_id: str,
         adapter: Adapter,
+        run_id: str,
+        id: str | None = None,
         config: dict[str, Any] | None = None,
         metadata: dict[str, Any] | None = None,
     ) -> None:
         """
         Initialize an environment.
-        
+
         Args:
-            run_id: ID of the run this environment belongs to
             adapter: Adapter for converting actions
+            run_id: ID of the run this environment belongs to
+            id: Optional ID of an existing environment
             config: Optional configuration parameters
             metadata: Optional metadata for the environment
         """
@@ -96,16 +115,14 @@ class Environment:
         self.config = config
         self.adapter = adapter
         self.metadata = metadata
-        # task_run_id is created when the environment is created (create_environment)
-        # or provided if env already exists.
         self.final_response: None | str = None
-        self.id = None
+        self.id = id
         self.vnc_url = None
 
     async def create_environment(self) -> str:
         """
         Initialize the environment and return the task_run_id.
-        
+
         Returns:
             str: The environment ID
         """
@@ -121,7 +138,7 @@ class Environment:
     async def get_vnc_url(self) -> str:
         """
         Get the VNC URL for the environment.
-        
+
         Returns:
             str: The VNC URL for remote viewing/control
         """
@@ -136,7 +153,7 @@ class Environment:
     async def get_env_state(self) -> str:
         """
         Get the state of the environment.
-        
+
         Returns:
             str: The current state (e.g., "running", "error")
         """
@@ -152,10 +169,10 @@ class Environment:
     ) -> tuple[Observation, float, bool, dict[str, Any]]:
         """
         Send action to environment and get result.
-        
+
         Args:
             action: The action to take, or None for no action
-            
+
         Returns:
             tuple: (observation, reward, terminated, info)
         """
@@ -181,10 +198,10 @@ class Environment:
     def translate_action(self, action: Any) -> list:
         """
         Translate action to the correct format.
-        
+
         Args:
             action: The action to translate
-            
+
         Returns:
             list: List of translated actions in the CLA format
         """
@@ -193,19 +210,19 @@ class Environment:
             return self.adapter.adapt_list(action)
         return [self.adapter.adapt(action)]
 
-    async def evaluate(self) -> float:
+    async def evaluate(self) -> RewardResponse:
         """
         Get final evaluation score.
-        
+
         Returns:
-            float: The evaluation score
+            RewardResponse: The evaluation response containing reward, logs, and possible error
         """
         data = await make_request(
             method="POST",
             url=f"{settings.base_url}/evaluation/{self.id}",
             api_key=settings.api_key,
         )
-        return data["reward"]
+        return RewardResponse(**data)
 
     async def close(self) -> None:
         """
@@ -217,16 +234,14 @@ class Environment:
             api_key=settings.api_key,
         )
 
-    async def reset(
-        self, task_id: str, metadata: dict[str, Any] | None = None
-    ) -> Observation:
+    async def reset(self, task_id: str, metadata: dict[str, Any] | None = None) -> Observation:
         """
         Reset the environment to the task.
-        
+
         Args:
             task_id: ID of the task to reset to
             metadata: Optional metadata for the reset
-            
+
         Returns:
             Observation: Initial observation for the task
         """
@@ -253,10 +268,11 @@ class Environment:
                 break
             await asyncio.sleep(10)
 
+
 class EvalSet:
     """
     Evaluation set containing tasks for benchmarking.
-    
+
     Attributes:
         id: Unique identifier for the evalset
         name: Human-readable name
@@ -268,10 +284,11 @@ class EvalSet:
         id: str,
         name: str,
         tasks: list[str] | None = None,
+        configs: dict[str, Any] | None = None,
     ) -> None:
         """
         Initialize an evaluation set.
-        
+
         Args:
             id: Unique identifier
             name: Human-readable name
@@ -280,11 +297,12 @@ class EvalSet:
         self.id = id
         self.name = name
         self.tasks = tasks or []
+        self.configs = configs or {}
 
     async def fetch_tasks(self) -> list[str]:
         """
         Fetch all tasks in this evalset from the API.
-        
+
         Returns:
             list[str]: List of task IDs
         """
@@ -293,5 +311,7 @@ class EvalSet:
             url=f"{settings.base_url}/evalsets/{self.id}/tasks",
             api_key=settings.api_key,
         )
+        # Extracts a list of task ids and list of config objects for the evalset
         self.tasks = data["tasks"]
+        self.configs = data["evalset"]
         return self.tasks

hud/gym.py

@@ -4,7 +4,7 @@ from __future__ import annotations
 class Gym:
     """
     Represents a simulation environment in the HUD system.
-    
+
     Attributes:
         id: Unique identifier for the gym
         name: Human-readable name of the gym
@@ -13,7 +13,7 @@ class Gym:
     def __init__(self, id: str, name: str) -> None:
         """
         Initialize a gym.
-        
+
         Args:
             id: Unique identifier
             name: Human-readable name

hud/run.py

@@ -19,21 +19,20 @@ if TYPE_CHECKING:
 class RunResponse(BaseModel):
     """
     Response model for run data from the API.
-    
+
     Attributes:
         id: Unique identifier for the run
         name: Human-readable name of the run
         gym: Dictionary containing gym information
         evalset: Dictionary containing evalset information
-        adapter: Dictionary containing adapter information
         config: Dictionary containing configuration parameters
         metadata: Dictionary containing metadata
     """
+
     id: str
     name: str
     gym: dict[str, Any]
     evalset: dict[str, Any]
-    adapter: dict[str, Any]
     config: dict[str, Any]
     metadata: dict[str, Any]
 
@@ -41,7 +40,7 @@ class RunResponse(BaseModel):
 class RunAnalyticsResponse(BaseModel):
     """
     Model for Run analytics data.
-    
+
     Attributes:
         id: Unique identifier for the run
         name: Human-readable name of the run
@@ -54,6 +53,7 @@ class RunAnalyticsResponse(BaseModel):
         created_at: When the run was created
         raw_data: Detailed data about tasks and environments
     """
+
     id: str
     name: str
     status_counts: dict[str, int]  # e.g. {"completed": 5, "running": 2, "error": 1}
@@ -69,7 +69,7 @@ class RunAnalyticsResponse(BaseModel):
 
     def __str__(self) -> str:
         return self.visualize()
-    
+
     def visualize(self) -> str:
         """
         Generate an ASCII bar chart visualization of run analytics.
@@ -82,9 +82,7 @@ class RunAnalyticsResponse(BaseModel):
         """
         max_width = 50
 
-        completion_rate = (
-            self.completion_rate if self.completion_rate is not None else 0
-        )
+        completion_rate = self.completion_rate if self.completion_rate is not None else 0
 
         result = [
             f"Run: {self.name} (ID: {self.id})",
@@ -123,7 +121,7 @@ class RunAnalyticsResponse(BaseModel):
 class Run:
     """
     A run represents a collection of tasks and environments.
-    
+
     This class provides methods to fetch task IDs, create environments,
     and access analytics for the run.
     """
@@ -140,7 +138,7 @@ class Run:
     ) -> None:
         """
         Initialize a run.
-        
+
         Args:
             id: Unique identifier
             name: Human-readable name
@@ -167,7 +165,7 @@ class Run:
     async def fetch_task_ids(self) -> list[str]:
         """
         Fetch task IDs for this run from the evalset.
-        
+
         Returns:
             list[str]: List of task IDs
         """
@@ -176,10 +174,10 @@ class Run:
     async def make(self, metadata: dict[str, Any] | None = None) -> Environment:
         """
         Create a new environment for this run.
-        
+
         Args:
             metadata: Metadata for the environment
-            
+
         Returns:
             Environment: The created environment
         """
@@ -197,7 +195,7 @@ class Run:
     async def get_analytics(self) -> RunAnalyticsResponse:
         """
         Get analytics for this run.
-        
+
         Returns:
             RunAnalyticsResponse: Analytics data including status counts,
                                 average score, and other metrics

hud/server/__init__.py

@@ -1,5 +1,5 @@
 from __future__ import annotations
 
-from .requests import RequestError, make_request, make_sync_request
+from .requests import RequestError, make_request
 
-__all__ = ["RequestError", "make_request", "make_sync_request"]
+__all__ = ["RequestError", "make_request"]

hud/server/requests.py

@@ -4,22 +4,27 @@ HTTP request utilities for the HUD API.
 
 from __future__ import annotations
 
+import asyncio
 import logging
 from typing import Any
 
 import httpx
 
+# Set up logger
 logger = logging.getLogger("hud.http")
+logger.setLevel(logging.DEBUG)
+
 
 class RequestError(Exception):
     """Custom exception for API request errors"""
+
     def __init__(
         self,
         message: str,
         status_code: int | None = None,
         response_text: str | None = None,
         response_json: dict[str, Any] | None = None,
-        response_headers: dict[str, str] | None = None
+        response_headers: dict[str, str] | None = None,
     ) -> None:
         self.message = message
         self.status_code = status_code
@@ -33,17 +38,22 @@ class RequestError(Exception):
 
         if self.status_code:
             parts.append(f"Status: {self.status_code}")
+
         if self.response_text:
             parts.append(f"Response Text: {self.response_text}")
+
         if self.response_json:
             parts.append(f"Response JSON: {self.response_json}")
+
         if self.response_headers:
             parts.append(f"Headers: {self.response_headers}")
 
         return " | ".join(parts)
 
     @classmethod
-    def from_http_error(cls, error: httpx.HTTPStatusError) -> RequestError:
+    def from_http_error(
+        cls, error: httpx.HTTPStatusError, context: str = ""
+    ) -> RequestError:
         """Create a RequestError from an HTTP error response"""
         response = error.response
         status_code = response.status_code
@@ -68,6 +78,10 @@ class RequestError(Exception):
             # Fallback to simple message if JSON parsing fails
             message = f"Request failed with status {status_code}"
 
+        # Add context if provided
+        if context:
+            message = f"{context}: {message}"
+
         # Log the error details
         logger.error(
             "HTTP error from HUD SDK: %s | URL: %s | Status: %s | Response: %s%s",
@@ -75,7 +89,7 @@ class RequestError(Exception):
             response.url,
             status_code,
             response_text[:500],
-            "..." if len(response_text) > 500 else ""
+            "..." if len(response_text) > 500 else "",
         )
 
         return cls(
@@ -86,61 +100,44 @@ class RequestError(Exception):
             response_headers=response_headers,
         )
 
-async def make_request(
-    method: str, url: str, json: Any | None = None, api_key: str | None = None
-) -> dict[str, Any]:
-    """
-    Make an asynchronous HTTP request to the HUD API.
-    
-    Args:
-        method: HTTP method (GET, POST, etc.)
-        url: Full URL for the request
-        json: Optional JSON serializable data
-        api_key: API key for authentication
-        
-    Returns:
-        dict: JSON response from the server
-        
-    Raises:
-        RequestError: If API key is missing or request fails
-    """
-    if not api_key:
-        raise RequestError("API key is required but not provided")
-
-    headers = {"Authorization": f"Bearer {api_key}"}
 
-    async with httpx.AsyncClient(timeout=240.0) as client:
-        try:
-            response = await client.request(
-                method=method, url=url, json=json, headers=headers
-            )
-            response.raise_for_status()
-            result = response.json()
-            return result
-        except httpx.HTTPStatusError as e:
-            raise RequestError.from_http_error(e) from None
-        except httpx.RequestError as e:
-            raise RequestError(f"Network error: {e!s}") from None
-        except Exception as e:
-            # Catch-all for unexpected errors
-            raise RequestError(f"Unexpected error: {e!s}") from None
+async def _handle_retry(
+    attempt: int, max_retries: int, retry_delay: float, url: str, error_msg: str
+) -> None:
+    """Helper function to handle retry logic and logging."""
+    retry_time = retry_delay * (2 ** (attempt - 1))  # Exponential backoff
+    logger.warning(
+        "%s from %s, retrying in %.2f seconds (attempt %d/%d)",
+        error_msg,
+        url,
+        retry_time,
+        attempt,
+        max_retries,
+    )
+    await asyncio.sleep(retry_time)
 
 
-def make_sync_request(
-    method: str, url: str, json: Any | None = None, api_key: str | None = None
+async def make_request(
+    method: str,
+    url: str,
+    json: Any | None = None,
+    api_key: str | None = None,
+    max_retries: int = 4,
+    retry_delay: float = 2.0,
 ) -> dict[str, Any]:
     """
-    Make a synchronous HTTP request to the HUD API.
-    
+    Make an asynchronous HTTP request to the HUD API.
+
     Args:
         method: HTTP method (GET, POST, etc.)
         url: Full URL for the request
         json: Optional JSON serializable data
         api_key: API key for authentication
-        
+        max_retries: Maximum number of retries
+        retry_delay: Delay between retries
     Returns:
         dict: JSON response from the server
-        
+
     Raises:
         RequestError: If API key is missing or request fails
     """
@@ -148,19 +145,49 @@ def make_sync_request(
         raise RequestError("API key is required but not provided")
 
     headers = {"Authorization": f"Bearer {api_key}"}
+    retry_status_codes = [502, 503, 504]
+    attempt = 0
+
+    while attempt <= max_retries:
+        attempt += 1
 
-    with httpx.Client(timeout=240.0) as client:
         try:
-            response = client.request(
-                method=method, url=url, json=json, headers=headers
-            )
+            async with httpx.AsyncClient(
+                timeout=240.0,
+                limits=httpx.Limits(
+                    max_connections=1000,
+                    max_keepalive_connections=1000,
+                    keepalive_expiry=10.0,
+                ),
+            ) as client:
+                response = await client.request(
+                    method=method, url=url, json=json, headers=headers
+                )
+
+            # Check if we got a retriable status code
+            if response.status_code in retry_status_codes and attempt <= max_retries:
+                await _handle_retry(
+                    attempt,
+                    max_retries,
+                    retry_delay,
+                    url,
+                    f"Received status {response.status_code}",
+                )
+                continue
+
             response.raise_for_status()
             result = response.json()
             return result
         except httpx.HTTPStatusError as e:
             raise RequestError.from_http_error(e) from None
         except httpx.RequestError as e:
-            raise RequestError(f"Network error: {e!s}") from None
+            if attempt <= max_retries:
+                await _handle_retry(
+                    attempt, max_retries, retry_delay, url, f"Network error: {e}"
+                )
+                continue
+            else:
+                raise RequestError(f"Network error: {e!s}") from None
         except Exception as e:
-            # Catch-all for unexpected errors
             raise RequestError(f"Unexpected error: {e!s}") from None
+    raise RequestError(f"Request failed after {max_retries} retries with unknown error")

hud/settings.py

@@ -7,32 +7,30 @@ from pydantic_settings import BaseSettings, SettingsConfigDict
 class Settings(BaseSettings):
     """
     Global settings for the HUD SDK.
-    
+
     This class manages configuration values loaded from environment variables
     and provides global access to settings throughout the application.
     """
-    model_config = SettingsConfigDict(
-        env_file=".env",
-        env_file_encoding="utf-8",
-        extra="allow"
-    )
+
+    model_config = SettingsConfigDict(env_file=".env", env_file_encoding="utf-8", extra="allow")
 
     base_url: str = Field(
         default="https://orchestrator.hud.live/hud-gym/api/v1",
         description="Base URL for the HUD API",
-        validation_alias="base_url"
+        validation_alias="base_url",
     )
 
     api_key: str | None = Field(
         default=None,
         description="API key for authentication with the HUD API",
-        validation_alias="HUD_API_KEY"
+        validation_alias="HUD_API_KEY",
     )
 
 
 # Create a singleton instance
 settings = Settings()
 
+
 # Add utility functions for backwards compatibility
 def get_settings() -> Settings:
     """Get the global settings instance."""

{hud_python-0.1.0b2.dist-info → hud_python-0.1.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hud-python
-Version: 0.1.0b2
+Version: 0.1.1
 Summary: SDK for the HUD evaluation platform.
 Project-URL: Homepage, https://github.com/Human-Data/hud-sdk
 Project-URL: Bug Tracker, https://github.com/Human-Data/hud-sdk/issues
@@ -57,9 +57,9 @@ Description-Content-Type: text/markdown
 
 # HUD
 
-A Python SDK for interacting with HUD environments and evaluation benchmarks for browser use and computer use models. Visit [hud.so](https://hud.so).
+A Python SDK for interacting with HUD environments and evaluation benchmarks for browser use and computer use models.
 
-> **Alpha Release Notice**: This SDK is currently in alpha status (v0.1.0-alpha). The API is evolving and may change in future releases as we gather feedback and improve functionality.
+> **Alpha Release Notice**: This SDK is currently in early release status. The API is evolving and may change in future releases as we gather feedback and improve functionality.
 
 [![PyPI version](https://img.shields.io/pypi/v/hud-python)](https://pypi.org/project/hud-python/)
 
@@ -70,13 +70,12 @@ A Python SDK for interacting with HUD environments and evaluation benchmarks for
 
 [RECOMMENDED] To set get started with an agent, see the [Claude Computer use example](https://github.com/Human-Data/hud-sdk/tree/main/examples).
 
-
-Otherwise, install the package with Python>=3.9:
+Install the package with Python>=3.9:
 ```bash
 pip install hud-python
 ```
 
-Make sure to setup your account [here](https://hud.so/settings) and add your API key to the environment variables:
+Make sure to setup your account with us (email founders@hud.so) and add your API key to the environment variables:
 ```bash
 HUD_API_KEY=<your-api-key>
 ```
@@ -95,7 +94,7 @@ async def main():
     evalset = await client.load_evalset(id="OSWorld-Ubuntu")
     
     # Create a run and environment
-    run = client.create_run(name="example-run", gym=gym, evalset=evalset)
+    run = await client.create_run(name="example-run", gym=gym, evalset=evalset)
     env = await run.make(metadata={"agent_id": "OSWORLD-1"})
     await env.wait_for_ready()
     
@@ -117,20 +116,9 @@ if __name__ == "__main__":
     asyncio.run(main())
 ```
 
-## Features
-
-- Connect to HUD evaluation environments
-- Run benchmarks across various tasks
-- Support for different agent adapters
-- Asynchronous API
-
 ## Documentation
 
-For comprehensive guides, examples, and API reference, visit:
-- [Getting Started](https://docs.hud.so/introduction)
-- [Installation](https://docs.hud.so/installation)
-- [API Reference](https://docs.hud.so/api-reference)
-- [Examples](https://docs.hud.so/examples)
+For comprehensive guides, examples, and API reference, visit [our docs](https://docs.hud.so/introduction)
 
 ## License
 

hud_python-0.1.1.dist-info/RECORD

@@ -0,0 +1,21 @@
+hud/__init__.py,sha256=8zbB2Wts-sHJ6pLPxwFDvM3Fo1RNT2r5m1H1NpMvyeA,414
+hud/client.py,sha256=7WHXTQhVK-T9Rj4ZooADE_c1pah5Bc1DJ9ZRqUyUnuQ,5724
+hud/environment.py,sha256=9r8eK3OVqr-wpPGlhnrpuDt-z6FIp3S3oukTZ7swN3o,8899
+hud/gym.py,sha256=aanBHtlsXrJwrFax9SbXWwk_By-X8wE3M9deS-E_s4c,463
+hud/run.py,sha256=5ukjuRNLjj5fczaWxpR_5NebFbQpoy8w81eRYy309Vg,6401
+hud/settings.py,sha256=1ScSac0ta03LkckkH2gi6SyKY2M7nr15vRGugo2C_xs,1015
+hud/adapters/__init__.py,sha256=y3H7yMl7rC-rrXG2WvePdSojoNFSui02eYTH17Xd7OY,87
+hud/adapters/claude/__init__.py,sha256=i7QEF-29FLb9qxp1eYtXs-adIk_tG54tL-9g6d3xodk,100
+hud/adapters/claude/adapter.py,sha256=ekYZixANKfx-4lENlXGaomh6Ecw4SRKtLWD5quGNWdM,5782
+hud/adapters/common/__init__.py,sha256=BjdZWJVs_AKtpFrt-tNsdQRjnz7D97DFEQirJ-r0mp8,118
+hud/adapters/common/adapter.py,sha256=SCtOuRjW5Szzd45LXCaqDEaKr2lhA-nIqSEMJ9KLsKI,5799
+hud/adapters/common/types.py,sha256=d9tIF06tjK7VCb-yBJ9epwHlXRHlObo9YWetrv33s8c,4511
+hud/server/__init__.py,sha256=VPrhyyqg3inge9J7BjcmDBNJRuvkCA9ZDXS_R5Q8ZtY,129
+hud/server/requests.py,sha256=pPPaMpwqmA1RyWwzQN1ybgAnbSHJLeeIaW6MJwhJYks,6052
+hud/utils/__init__.py,sha256=0m8klSLnMLeIJT23ipBXfFACk4hNWPsA6ZNqZDpv6oY,99
+hud/utils/config.py,sha256=dze0BGE4q14omjj9822kL9BeiIgWQvJyuU29A2wa1SE,193
+hud/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+hud_python-0.1.1.dist-info/METADATA,sha256=rH6sSZl4J7sfQdUp2QxNRUp9scqRMG2TxI390KD3Lr4,4785
+hud_python-0.1.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+hud_python-0.1.1.dist-info/licenses/LICENSE,sha256=IVdfcZ8xq5apYGJS5GzRLLbm9r03Aecxd03isi-3P9k,1075
+hud_python-0.1.1.dist-info/RECORD,,

hud_python-0.1.0b2.dist-info/RECORD

@@ -1,21 +0,0 @@
-hud/__init__.py,sha256=GmX-LujM2oZR6_tP_mOW09BY8HeK41lLF-P0sMW_1pY,416
-hud/client.py,sha256=ztWPiAJyJUdJxdxGqDmsQnVK-_jccinWQUUXmq0OOmY,5843
-hud/environment.py,sha256=R-t-21V0gveuHL6LlVLnBp0gYGm8tn5FbLcq_rRlH9g,8587
-hud/gym.py,sha256=dKmf0Ol0-XyLhji034pF_5dXnhW1IgIr-dJUg4KfslE,475
-hud/run.py,sha256=_K7POPjJyqcJ_DVLAO7hRmvLUcg9gg2KrLHw_26DB9I,6570
-hud/settings.py,sha256=FbZHI1q6bDHe7Awl32SDPb-syqtkLI3C7gIIXuMXCiQ,1045
-hud/adapters/__init__.py,sha256=y3H7yMl7rC-rrXG2WvePdSojoNFSui02eYTH17Xd7OY,87
-hud/adapters/claude/__init__.py,sha256=GsMxaBL5ZuKV6-jJsLfw23n_Ml9e88SXIddYDGkIUKE,101
-hud/adapters/claude/adapter.py,sha256=oi2lvO42g7i-L151tIWIGQGA80skcYRwzQ52-0f2OpA,4840
-hud/adapters/common/__init__.py,sha256=BjdZWJVs_AKtpFrt-tNsdQRjnz7D97DFEQirJ-r0mp8,118
-hud/adapters/common/adapter.py,sha256=SCtOuRjW5Szzd45LXCaqDEaKr2lhA-nIqSEMJ9KLsKI,5799
-hud/adapters/common/types.py,sha256=LlWxH9sWucYgnIv6DKrgqToh3k7Bu-xdTxNFU4L8Xg8,1962
-hud/server/__init__.py,sha256=HeIXBGb-bxtq3xF20jP4IrOy77PlsqhClOf3bZ9wrwI,169
-hud/server/requests.py,sha256=M_pK1oCd4QjIE0yguD6iaybJ_mempOWDQYEpdOkophU,5522
-hud/utils/__init__.py,sha256=0m8klSLnMLeIJT23ipBXfFACk4hNWPsA6ZNqZDpv6oY,99
-hud/utils/config.py,sha256=dze0BGE4q14omjj9822kL9BeiIgWQvJyuU29A2wa1SE,193
-hud/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-hud_python-0.1.0b2.dist-info/METADATA,sha256=5skHs5IfSJP4DQAGzpuo_yjO7l65XxHIPWGitHQt0Ug,5140
-hud_python-0.1.0b2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-hud_python-0.1.0b2.dist-info/licenses/LICENSE,sha256=IVdfcZ8xq5apYGJS5GzRLLbm9r03Aecxd03isi-3P9k,1075
-hud_python-0.1.0b2.dist-info/RECORD,,