What You’ll Learn
- Sending prompts to the OpenAI chat API (
gpt-4o-mini) to generate executable Python code
- Stripping markdown code fences from LLM responses before execution
- Writing generated code into a Declaw sandbox filesystem
- Executing the code securely with
sbx.commands.run()
- Graceful demo mode when no API key is configured
Prerequisites
- Declaw instance running and
DECLAW_API_KEY / DECLAW_DOMAIN set
OPENAI_API_KEY (optional — the example runs in demo mode without it)
pip install declaw python-dotenv openai
npm install @declaw/sdk dotenv openai
Code Walkthrough
1. Ask GPT-4o-mini to generate Python code
The system prompt instructs the model to return only code — no markdown, no explanation.import openai
from declaw import Sandbox
client = openai.OpenAI()
response = client.chat.completions.create(
model="gpt-4o-mini",
messages=[
{
"role": "system",
"content": (
"You are a Python code interpreter. When asked a question, "
"respond ONLY with Python code that computes the answer and "
"prints it. No markdown, no explanation, just code."
),
},
{"role": "user", "content": "What are the first 20 prime numbers?"},
],
temperature=0,
)
2. Strip code fences and execute in a sandbox
Models occasionally wrap code in triple-backtick fences even when asked not to. Strip them before writing to disk:def strip_code_fences(code: str) -> str:
code = code.strip()
if code.startswith("```"):
code = "\n".join(code.split("\n")[1:])
if code.endswith("```"):
code = "\n".join(code.split("\n")[:-1])
return code.strip()
code = strip_code_fences(response.choices[0].message.content or "")
sbx = Sandbox.create(template="python", timeout=300)
try:
sbx.files.write("/tmp/solution.py", code)
result = sbx.commands.run("python3 /tmp/solution.py", timeout=30)
print(result.stdout)
print(result.exit_code)
finally:
sbx.kill()
3. Demo mode (no API key needed)
When OPENAI_API_KEY is not set the example runs pre-written code to show the Declaw integration:code = """\
def primes(n):
result = []
candidate = 2
while len(result) < n:
if all(candidate % p != 0 for p in result):
result.append(candidate)
candidate += 1
return result
print("First 20 primes:", primes(20))
print("Factorial of 15:", __import__('math').factorial(15))
"""
sbx = Sandbox.create(template="python", timeout=300)
try:
sbx.files.write("/tmp/demo.py", code)
result = sbx.commands.run("python3 /tmp/demo.py", timeout=30)
print(result.stdout)
finally:
sbx.kill()
1. Ask GPT-4o-mini to generate Python code
import { Sandbox } from "@declaw/sdk";
import OpenAI from "openai";
const client = new OpenAI();
const response = await client.chat.completions.create({
model: "gpt-4o-mini",
messages: [
{
role: "system",
content:
"You are a Python code interpreter. When asked a question, respond ONLY with Python code that computes the answer and prints it. No markdown, no explanation, just code.",
},
{ role: "user", content: "What are the first 20 prime numbers?" },
],
temperature: 0,
});
2. Strip code fences and execute in a sandbox
function stripCodeFences(code: string): string {
let cleaned = code.trim();
if (cleaned.startsWith("```")) {
cleaned = cleaned.split("\n").slice(1).join("\n");
}
if (cleaned.endsWith("```")) {
cleaned = cleaned.split("\n").slice(0, -1).join("\n");
}
return cleaned.trim();
}
const code = stripCodeFences(response.choices[0].message.content || "");
const sbx = await Sandbox.create({ template: "python", timeout: 300 });
try {
await sbx.files.write("/tmp/solution.py", code);
const result = await sbx.commands.run("python3 /tmp/solution.py", {
timeout: 30,
});
console.log(result.stdout);
console.log(result.exitCode);
} finally {
await sbx.kill();
}
3. Demo mode (no API key needed)
const code = `
def primes(n):
result = []
candidate = 2
while len(result) < n:
if all(candidate % p != 0 for p in result):
result.append(candidate)
candidate += 1
return result
print("First 20 primes:", primes(20))
print("Factorial of 15:", __import__('math').factorial(15))
`.trim();
const sbx = await Sandbox.create({ template: "python", timeout: 300 });
try {
await sbx.files.write("/tmp/demo.py", code);
const result = await sbx.commands.run("python3 /tmp/demo.py", { timeout: 30 });
console.log(result.stdout);
} finally {
await sbx.kill();
}
Expected Output
============================================================
OpenAI Code Interpreter with Declaw Sandbox
============================================================
--- Question: What are the first 20 prime numbers? ---
Generated code:
def primes(n): ...
Result:
stdout: First 20 primes: [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71]
stderr:
exit_code: 0
============================================================
Done!
============================================================
In demo mode (no OPENAI_API_KEY), the output shows pre-written code results: the first 20 primes and factorial(15) = 1307674368000.
