# Run AI Apps with the inference.sh CLI

Using many hosted AI models from a terminal requires different app names, inputs, and task flows. This skill provides focused inference.sh CLI guidance.

## Install

```bash
npx skillstore add inferen-sh/infsh-cli
```

## Metadata

- Status: approved
- Slug: inferen-sh-infsh-cli
- Skillstore revision: r2
- Version status: missing
- Tree hash: e3f0051c7782dc25693866f11341633162524ceeb361d5736c353ee26381e845
- Author: inferen-sh
- GitHub username: inferen-sh
- License: MIT
- Repository: https://github.com/inferen-sh/skills/tree/main/tools/infsh-cli/
- Ref: 4121de961d1b6f2ffca856260e239505c302452c
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: blocked
- Manual install advisory: allowed\_with\_warning
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: network, env\_access, filesystem, external\_commands
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/inferen-sh-infsh-cli
- Manifest: https://skillstore.pages.dev/api/skills/inferen-sh-infsh-cli/manifest

## Capabilities

- Search and filter the public inference.sh app catalog.
- Inspect app details and generate sample input files.
- Run image, video, language, search, audio, and 3D apps.
- Upload user-selected local media as app inputs.
- Submit long-running tasks and retrieve their status or results.
- Guide local testing and deployment of inference.sh apps.

## Use Cases

- Create Media Prototypes: Find and run image or video apps, then monitor long-running generation tasks.
- Evaluate Hosted Models: Compare language, search, speech, and media apps through repeatable CLI inputs.
- Automate App Operations: Inspect, test, deploy, and invoke inference.sh apps from development or CI workflows.

## Prompt Templates

### Find an App

```
Find inference.sh apps for [task]. Show the most relevant choices and explain their main differences before running anything.
```

### Prepare a Safe Run

```
Inspect [app-name], create a sample input for [goal], and show the planned command. Ask before submitting the task.
```

### Run and Track a Task

```
Run [app-name] with [input-file] after confirming any local uploads. Return the task identifier, status, and final output location.
```

### Build a Reproducible Workflow

```
Design a version-pinned inference.sh workflow for [project]. Validate schemas, protect secrets, confirm uploads, monitor tasks, and document failure recovery.
```

## Limitations

- Requires the external belt CLI, an inference.sh account, network access, and available credits.
- App names, model availability, schemas, prices, and outputs can change outside this skill.
- Local files may be uploaded to a hosted service when their paths are used as inputs.
- Installation and social media actions require careful user review because they have external effects.

## Best Practices

- Inspect each app schema and generate a sample input before the first run.
- Pin app versions and preserve task identifiers for reproducible workflows.
- Confirm costs, local file uploads, credentials, and external account changes before execution.

## Anti Patterns

- Do not pipe remote installer output directly into a shell.
- Do not place API keys in prompts, command history, source files, or saved examples.
- Do not publish social content or upload local files without explicit user approval.

## Security Audit

- Audited at: 2026-08-06T10:31:07.13\+00:00
- Summary: The documentation contains repeated remote installer commands that pipe live content to a shell, plus an unsafe manifest-driven download command. It also enables local file uploads and public social media actions without explicit consent controls; most other alerts are Markdown or documentation false positives.

## Stats

- Views: 235
- Downloads: 42
- Favorites: 0
- Popularity score: 0