code-interpreter

Code Interpreter with Workers AI

A Cloudflare Worker that gives the gpt-oss model on Workers AI the ability to execute Python code using the Cloudflare Sandbox SDK.

Features

• Workers AI Integration: Uses @cf/openai/gpt-oss-120b via the workers-ai-provider package
• Vercel AI SDK: Leverages generateText() and tool() for clean function calling
• Sandbox Execution: Python code runs in isolated Cloudflare Sandbox containers

How It Works

1. User sends a prompt to the /run endpoint
2. GPT-OSS receives the prompt with an execute_python tool
3. Model decides if Python execution is needed
4. Code runs in an isolated Cloudflare Sandbox container
5. Results are sent back to the model for final response

API Endpoint

POST /run
Content-Type: application/json

{
  "input": "Your prompt here"
}

Example Usage

# Simple calculation
curl -X POST http://localhost:8787/run \
  -H "Content-Type: application/json" \
  -d '{"input": "Calculate 5 factorial using Python"}'

# Execute specific code
curl -X POST http://localhost:8787/run \
  -H "Content-Type: application/json" \
  -d '{"input": "Execute this Python: print(sum(range(1, 101)))"}'

# Complex operations
curl -X POST http://localhost:8787/run \
  -H "Content-Type: application/json" \
  -d '{"input": "Use Python to find all prime numbers under 20"}'

Setup

1. From the project root:

npm install
npm run build

2. Run locally:

cd examples/code-interpreter
npm run dev

Note: First run builds the Docker container (2-3 minutes). Subsequent runs are much faster.

Deploy

npx wrangler deploy

Wait for provisioning: After first deployment, wait 2-3 minutes before making requests.