py-sandbox 0.2.0

Execute untrusted Python code safely with caching and remote loading

py-sandbox

Execute untrusted Python code safely with isolation, caching, and integrity verification.

Installation

pip install py-sandbox

Quick Start

from py_sandbox import run

result = run('''
x = [1, 2, 3, 4, 5]
result = sum(x) * 2
print(f"Calculated: {result}")
''')

print(result['stdout'])   # "Calculated: 30"
print(result['result'])   # 30
print(result['success'])  # True

Security

Dangerous operations are automatically blocked:

result = run("import os; os.system('rm -rf /')"  )
print(result['success'])  # False
print(result['stderr'])   # "Import 'os' blocked for security"

Static Analysis

Check code before running:

from py_sandbox import check

analysis = check("import subprocess; subprocess.call(['ls'])")
print(analysis['safe'])      # False
print(analysis['warnings'])  # ['Subprocess access attempt']

Caching

from py_sandbox import cache

cache.set("key", {"data": 123}, ttl=3600)
value = cache.get("key")
cache.delete("key")
cache.clear()
print(cache.stats())

Integrity Verification

from py_sandbox import verify

hash = verify.sign("my content")
assert verify.check("my content", hash)

Remote Loading

Load modules from authenticated endpoints:

from py_sandbox import remote

remote.configure(
    endpoint="https://your-server.com/modules",
    api_key="your-key"
)

module = remote.load("module_name")

Environment Variables

export PY_SANDBOX_ENDPOINT="https://your-server.com/modules"
export PY_SANDBOX_KEY="your-key"

License

MIT