flux-best-practices

Generates optimized prompts for BFL FLUX image generation and editing models.

Updated Jul 16, 2026
One-click install
npx skills add https://github.com/X-manist/Cohmira --skill flux-best-practices-x-manist
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: flux-best-practices
Source: https://github.com/X-manist/Cohmira/tree/main/src/builtin-plugins/openmontage/.agents/skills/flux-best-practices
Command: npx skills add https://github.com/X-manist/Cohmira --skill flux-best-practices-x-manist

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Writing effective prompts for FLUX image generation models requires knowing model-specific rules, such as the lack of negative prompt support, lighting emphasis, and per-model prompting styles. This Skill codifies those rules so generated prompts produce accurate, high-quality images on the first attempts. ## Core Features & Use Cases - Prompt Construction Guidance: Applies the subject-action-style-context-lighting-technical formula with positive-only phrasing, since FLUX does not support negative prompts. - Model Selection: Maps use cases (speed, quality, typography, editing, local development) to the right FLUX.2 or FLUX.1 model with cost and reference-image limits. - Advanced Techniques: Covers JSON structured prompting, hex color specification, typography rendering, and multi-reference image editing. - Use Case: A marketing designer needs a product hero image with exact brand colors and readable headline text; the Skill selects FLUX.2 [flex], quotes the text, and specifies hex colors with descriptive names. ## Quick Start Use the flux-best-practices skill to write a FLUX.2 prompt for a golden-hour product photo of headphones with brand color #FF6B6B.

Frequently Asked Questions about flux-best-practices

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write effective prompts for FLUX image generation?

Build prompts using the formula: subject, action, style, context, lighting, and technical details. Write descriptive prose of 30-80 words, front-load important elements, and always specify lighting since it has the greatest impact on image quality.

Which FLUX model should I use for my task?

Use FLUX.2 [klein] for speed, [max] for highest quality, [pro] for balanced production work, [flex] for typography, and [dev] for free local generation. All FLUX.2 models support image editing via reference images, so FLUX.1 Kontext is generally not needed.

Does FLUX support negative prompts?

No, FLUX does not support negative prompts. Instead of describing what to exclude, describe the positive alternative: use 'empty' or 'deserted' instead of 'no people', and 'natural skin' instead of 'no makeup'.

How do I edit existing images with FLUX.2?

Pass the source image URL directly to the input_image parameter and describe the desired change in the prompt. Be explicit about what to preserve, such as composition or style, and break dramatic transformations into sequential smaller edits.

Can FLUX render accurate text and brand colors in images?

Yes. Use FLUX.2 [flex] for typography, always quote exact text, and specify font style and placement. For colors, provide hex codes in #RRGGBB format paired with descriptive color names, limiting palettes to 3-5 colors.

How many reference images can FLUX models accept?

FLUX.2 [klein] accepts up to 4 reference images, [pro] and [flex] up to 8, and [max] up to 8 via API or 10 in the playground. Reference images by number in the prompt for precise multi-image compositions.