Metadata-Version: 2.4
Name: inspect_sandboxes
Version: 0.4.0
Summary: Collection of sandboxes for Inspect AI
Project-URL: Source Code, https://github.com/meridianlabs-ai/inspect_sandboxes
Project-URL: Issue Tracker, https://github.com/meridianlabs-ai/inspect_sandboxes/issues
Project-URL: Documentation, https://meridianlabs-ai.github.io/inspect_sandboxes/
Author: Meridian Labs
License: MIT License
License-File: LICENSE
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Requires-Python: >=3.12
Requires-Dist: daytona-sdk>=0.180.0
Requires-Dist: e2b>=2.20
Requires-Dist: inspect-ai>=0.3.211
Requires-Dist: modal>=1.3.3
Requires-Dist: tenacity>=9.1.4
Description-Content-Type: text/markdown

# Inspect Sandboxes

The `inspect-sandboxes` Python package provides cloud sandbox environments for [Inspect AI](https://inspect.aisi.org.uk/) evaluations. Each sample runs in a managed sandbox on your provider of choice (Daytona, E2B, or Modal) while the Inspect process runs locally on your machine.

This lets you run evaluations without a local Docker daemon or your own Kubernetes cluster, and scale to many concurrent samples beyond what fits on a single machine.

## Documentation

See the [documentation](https://meridianlabs-ai.github.io/inspect_sandboxes/) for installation, a provider quick start, and full configuration options for each provider.

## Development

```bash
# Install dependencies
make install

# Run tests (skips integration tests)
make test

# Run all tests including integration tests
make test-all

# Run type checking and linting
make check
```
