Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.openreward.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

You can develop and test OpenReward environments locally before deploying to production. This enables:
  • Rapid Iteration: Test changes instantly without waiting for deployments
  • Debugging: Use debuggers and inspect environment behavior
  • Offline Development: Work without internet connection
  • Cost Savings: No cloud resources used during development

​
Quick Start: Testing Environments Locally

​
1. Install Dependencies

pip install openreward

​
2. Write Your Environment Server

# server.py
from openreward.environments import Environment, Server, Split, TextBlock, ToolOutput, tool
from pydantic import BaseModel

class AnswerParams(BaseModel):
    answer: str

class MyEnvironment(Environment):
    """A simple math environment"""

    @classmethod
    def list_tasks(cls, split: str):
        return [
            {"id": "task-1", "problem": "What is 2+2?", "answer": 4},
            {"id": "task-2", "problem": "What is 5*3?", "answer": 15},
        ]

    @classmethod
    def list_splits(cls):
        return [Split(name="train", type="train"), Split(name="test", type="test")]

    def get_prompt(self):
        return [TextBlock(type="text", text=self.task_spec["problem"])]

    @tool
    def answer(self, params: AnswerParams) -> ToolOutput:
        """Submit your answer"""
        correct = str(self.task_spec["answer"]) == params.answer
        return ToolOutput(
            blocks=[TextBlock(type="text", text="Correct!" if correct else "Wrong!")],
            reward=1.0 if correct else 0.0,
            finished=True
        )

if __name__ == "__main__":
    Server([MyEnvironment]).run()

​
3. Run the Server

python server.py
Output: 
INFO:     Started server process [12345]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8080

​
4. Test the Environment

# test_environment.py
from openreward import OpenReward

or_client = OpenReward(base_url="http://localhost:8080") # you must point towards the environment if running locally

# Connect to local server (no namespace = localhost)
environment = or_client.environments.get(name="myenvironment")

# List tasks
tasks = environment.list_tasks(split="train")
print(f"Found {len(tasks)} tasks")

# Create session and test tool
with environment.session(task=tasks[0]) as session:
    prompt = session.get_prompt()
    print(f"Prompt: {prompt[0].text}")

    result = session.call_tool("answer", {"answer": "4"})
    print(f"Result: {result.blocks[0].text}")
    print(f"Reward: {result.reward}")
Alternatively, if you don’t want the whole client to point to http://localhost:8080 just change the base_url when getting the environment, like this: 
environment = or_client.environments.get(name="myenvironment", base_url="http://localhost:8080")
Run: 
python test_environment.py
Output: 
Found 2 tasks
Prompt: What is 2+2?
Result: Correct!
Reward: 1.0

​
Local Development with Docker

​
Build and Run Locally

# Build Docker image
docker build -t my-environment:local .

# Run container
docker run -p 8080:8080 my-environment:local

# Test from another terminal
python test_environment.py

​
Next Steps

Your First Environment

Complete tutorial for building an environment

GitHub Deployment

Deploy your local environment to OpenReward

Secret Management

Manage secrets in local and production environments

Environments

Understand environment architecture