# Run Python Tasks in a Hosted Sandbox

Local environments may lack the libraries or resources required for Python tasks. This skill sends workloads to a hosted CPU sandbox with common packages.

## Install

```bash
npx skillstore add inferen-sh/python-executor
```

## Metadata

- Status: approved
- Slug: inferen-sh-python-executor
- Skillstore revision: r2
- Version status: missing
- Tree hash: e7dfc159ab8ec1467df6a41d7d9453591d31715785fa4a066460349610403e6f
- Author: inferen-sh
- GitHub username: inferen-sh
- License: MIT
- Repository: https://github.com/inferen-sh/skills/tree/main/tools/utilities/python-executor/
- Ref: 4121de961d1b6f2ffca856260e239505c302452c
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 50
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/inferen-sh-python-executor
- Manifest: https://skillstore.pages.dev/api/skills/inferen-sh-python-executor/manifest

## Capabilities

- Runs Python 3.10 code through the infsh/python-executor application.
- Supports execution timeouts from 1 to 300 seconds.
- Provides common libraries for data, web, image, video, audio, 3D, and document tasks.
- Returns files written to the outputs directory.
- Offers default 8 GB and high-memory 16 GB CPU variants.

## Use Cases

- Analyze Business Data: Clean tabular data, calculate summaries, create charts, and return reusable CSV or image files.
- Create Media Assets: Generate or transform images, videos, audio, PDFs, and 3D models with installed Python libraries.
- Automate Public Data Tasks: Call approved APIs, parse public web pages, and format results without configuring a local Python environment.

## Prompt Templates

### Calculate Simple Statistics

```
Use python-executor to calculate summary statistics for these values: {values}. Return a short table and explain any outliers.
```

### Clean and Chart Data

```
Use python-executor with pandas to analyze {dataset_description}. Save cleaned data to outputs/cleaned.csv and a chart to outputs/chart.png.
```

### Extract Public Web Data

```
Use python-executor to fetch {public_url}, parse {fields}, and return a CSV. Respect access rules and do not send credentials.
```

### Run a High-Memory Workflow

```
Use the high-memory python-executor variant for {task}. Validate inputs, set an appropriate timeout, save artifacts under outputs, and summarize verification results.
```

## Limitations

- Execution requires the belt CLI, an authenticated inference.sh account, and network access.
- The environment is CPU-only and does not provide GPU machine-learning libraries.
- Workloads are non-interactive, so charts and media must be saved as files.
- Code and embedded data are processed by a third-party hosted service.

## Best Practices

- Review generated Python and external destinations before approving execution.
- Use explicit timeouts and choose the high-memory variant only when the workload requires it.
- Write every required artifact under outputs and verify file contents before presenting results.

## Anti Patterns

- Do not submit secrets, private datasets, or proprietary source without authorization.
- Do not use the sandbox for interactive applications or GPU-dependent workloads.
- Do not scrape restricted sites or call unapproved endpoints.

## Security Audit

- Audited at: 2026-08-06T10:55:16.929\+00:00
- Summary: Most external-command findings are false positives caused by Markdown backticks, fenced examples, or inline library names. Real risks remain: the skill runs belt and npx commands, submits generated Python to a hosted service, permits outbound HTTP in examples, and grants a broad Bash\(belt \*\) tool scope.

## Stats

- Views: 138
- Downloads: 9
- Favorites: 0
- Popularity score: 0