ARY / getting_started.md
harshitmahour360's picture
Initial commit for stage 3
e331bc2
|
Raw History Blame Contribute Delete
2.83 kB

Getting Started with SupportDesk OpenEnv

Welcome to the SupportDesk OpenEnv project! This guide will take you from a fresh clone to a fully running AI agent evaluation system in under 5 minutes.


πŸ“‹ Prerequisites

  • Python 3.11 or higher
  • (Optional) Docker

πŸ› οΈ Step 1: Environment Setup

First, create a virtual environment and install the package in "editable" mode. This ensures that any changes you make to the source code are applied immediately.

# Create the virtual environment
python -m venv venv

# Activate the environment (Windows)
.\venv\Scripts\activate

# Install dependencies
pip install -e .
pip install python-dotenv

πŸ”‘ Step 2: Configuration

The system uses environment variables for API keys and model settings. Create a file named .env in the root directory:

# .env file
API_BASE_URL=https://api-inference.huggingface.co/v1
MODEL_NAME=HuggingFaceH4/zephyr-7b-beta
HF_TOKEN=your_hugging_face_token_here

πŸ–₯️ Step 3: Running the Simulation Server

The simulation environment is a FastAPI web server. You must have this running before the AI agent can play the scenarios.

uvicorn supportdesk_env.server.app:app --host 127.0.0.1 --port 7860

Note: Keep this terminal window open. If you see "Application startup complete," the server is ready.


πŸ€– Step 4: Running the AI Agent

Open a separate terminal window to run the agent. The agent will read the scenarios from the server and attempt to solve them using your AI model.

Complete Run (All Tasks)

python inference.py

Run a Single Task

python inference.py --tasks login_lockout

Dry Run (Offline Mode)

Test the whole system without using any AI credits:

python inference.py --offline

πŸ§ͺ Step 5: Verification & Testing

Automated Unit Tests

Run the internal logic tests (checks the grader and environment mechanics):

pytest

Manual API check

Check if the server is healthy using curl:

curl http://127.0.0.1:7860/health

Smoke Test Script

We have included a specific script to verify the end-to-end API connection:

python smoke_test.py

🐳 Step 6: Docker Deployment (Optional)

If you want to run the environment inside a container:

# Build the image
docker build -t supportdesk-openenv .

# Run the container
docker run -p 7860:7860 supportdesk-openenv

πŸ“ Project Structure Quick-Map

  • supportdesk_env/: The core simulation logic.
  • task_bank.py: Where the customer scenarios are defined.
  • inference.py: The main entry point for the AI agent.
  • app.py: The FastAPI server entry point.
  • grader.py: Logic that scores the AI's performance.