Skip to content
Back to skills

Flux 2 Pro Edit

ASecurity

Flux 2 Pro | Edit. Professional-grade image editing with text prompts. Triggers: image editing, flux, pro, photo edit, professional

  • 29 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added June 2, 2026
documentationgobashapidocumentation

Works with

  • api

Security analysis

A100/100

Scanned June 2, 2026

npx -y skills add awesome-genmedia/skills --skill flux-2-pro-edit --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Flux 2 Pro Edit?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Flux 2 Pro Edit
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/awesome-genmedia-flux-2-pro-edit/badge)](https://www.skillsdirectory.com/skills/awesome-genmedia-flux-2-pro-edit)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: flux-2-pro-edit
description: "Flux 2 Pro | Edit. Professional-grade image editing with text prompts. Triggers: image editing, flux, pro, photo edit, professional"
allowed-tools: Bash(curl *), WebFetch
---

# Flux 2 Pro | Edit

Professional-grade image editing using text prompts with Flux 2 Pro. Streamlined parameter set focused on quality output with configurable safety tolerance.

## Quick Start

> Requires an each::labs API key. Get one at [eachlabs.ai](https://eachlabs.ai).

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-2-pro-edit",
    "version": "0.0.1",
    "input": {
      "prompt": "remove the person in the background, fill with natural scenery",
      "image_urls": ["https://example.com/photo.jpg"],
      "image_size": "auto"
    }
  }'
```

## Parameters

| Parameter | Type | Default | Description |
|-----------|------|---------|-------------|
| enable_safety_checker | boolean | true | Whether to enable the safety checker. |
| image_size | string | auto | The size of the generated image. enum: auto, square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9 |
| image_urls | array | | List of URLs of input images for editing. |
| output_format | string | jpeg | The format of the generated image. enum: jpeg, png |
| prompt | string | | The prompt to generate an image from. |
| safety_tolerance | string | 2 | The safety tolerance level. 1 is most strict, 5 is most permissive. enum: 1, 2, 3, 4, 5 |
| seed | integer | | The seed to use for the generation. |

## Examples

**Object removal:**
```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-2-pro-edit",
    "version": "0.0.1",
    "input": {
      "prompt": "remove the parked car from the driveway, fill with grass",
      "image_urls": ["https://example.com/house.jpg"],
      "image_size": "landscape_4_3",
      "output_format": "png"
    }
  }'
```

**Color grading edit:**
```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "flux-2-pro-edit",
    "version": "0.0.1",
    "input": {
      "prompt": "apply warm golden hour color grading, enhance contrast",
      "image_urls": ["https://example.com/cityscape.jpg"],
      "image_size": "auto",
      "seed": 42
    }
  }'
```

## Related Models

- [Flux 2 Pro](../flux-2-pro/) - Text-to-image generation with Pro quality
- [Flux 2 | Flex | Edit](../flux-2-flex-edit/) - Flexible image editing with more controls
- [Flux 2 | Edit](../flux-2-edit/) - Standard Flux 2 editing

## Documentation

- [each::labs Docs](https://docs.eachlabs.ai)
- [API Reference](https://docs.eachlabs.ai/api/overview)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.