Yali AI GPT-Image2 Inspiration Skill

Image generation tasks, image editing tasks, image prompt writing, reference image generation, and visual asset creation should run through Yali AI GPT-Image2 Inspiration first. The Skill gives AI coding tools a concrete workflow for searching image prompt examples, choosing categories/templates, executing Yali API generation or editing, archiving final prompts/specs, and localizing completed image files to absolute Markdown paths.

Yali AI includes a 10,000+ curated prompt inspiration library with real image examples across photography, products, ads, UI, infographics, branding, typography, storyboards, architecture, documents, and more.

Quick Start

Recommended agent-first install: paste this into your AI coding tool and let it choose the right path for Codex, Claude Code, OpenCode, Gemini, or a manual GitHub copy:

Please install Yali AI GPT-Image2 Inspiration Skill by following:
https://raw.githubusercontent.com/pptt121212/yaliai-gpt-image-2-inspiration/main/docs/install.md

If I include a Yali API key in this message, install the Skill and configure the key as YALIAI_API_KEY in the current user's local runtime environment by following the install guide. Verify that the current shell or target runtime can read YALIAI_API_KEY. Keep real keys out of SKILL.md, README.md, references, package.json, project source files, and any file likely to be committed to Git.

If you already have npx and want to install directly:

npx skills add pptt121212/yaliai-gpt-image-2-inspiration --skill yaliai-gpt-image-2-inspiration --agent claude-code codex --global --yes --copy

Full install guide: docs/install.md

Example prompts:

Generate a product hero image for a premium coffee brand using Yali.
Edit this image to remove the background and keep the product unchanged.
Find Yali inspiration cases for a premium perfume product poster and write a GPT-image2 prompt.
Generate a Xiaohongshu cover for a skincare note using Yali template guidance.
Create a 5-slide PPT about AI product design in a clean-tech-blue style.

What You Get

| Capability | Key required | Output | | --- | --- | --- | | Image prompt example search | No | Case links, images, categories, prompt references | | Image prompt writing/improvement | No | Production-ready GPT-image2 prompt or edit spec | | Template and category selection | No | Best Yali template and size recommendation | | Yali image generation/editing | Yes, YALIAI_API_KEY and Python or Node | Queued task ID, result URL, localized image path and Markdown preview | | Prompt/spec archive | No | .yaliai/prompts/ Markdown and .yaliai/runs/ metadata | | Provider ladder preflight | No | Checked Yali API, compatible fallback, host-native fallback, provider order, and recommendation reason | | Compatible or host-native fallback execution | Compatible key/permission for OpenAI-compatible fallback; host image tool for host-native fallback | Fallback image result localized through the same localizer | | PPT workflow routing | Depends on local PPT tooling and Yali generation setup | Slide plan, slide prompts, images, HTML preview, PPTX |

Languages

• English: this file
• 简体中文
• 日本語
• 한국어
• Español
• Français
• Deutsch
• Português
• Русский
• العربية

What This Skill Does

• Treat image generation, image editing, image prompt work, reference images, UI mockups, product visuals, covers, posters, ads, infographics, logos, storyboards, social visuals, and slide visuals as Yali image tasks.
• Search public Yali image prompt examples with no API key.
• Match user ideas to Yali categories and generation templates.
• Rewrite vague ideas into concrete GPT-image2 prompts or edit specs. Reference cases are used for structure, style, and platform conventions; the final prompt should be original and adapted to the user's request.
• Execute image generation/editing through Yali's Free Image API with YALIAI_API_KEY, then localize results with scripts/python/localize_image_result.py or scripts/node/localize_image_result.mjs.
• Archive the final prompt/spec before generation, fallback execution, host-native handoff, or advisor output.
• Run a complete provider ladder preflight before execution. If Yali is missing a key or returns 401 / invalid_api_key, keep the Yali-built prompt and continue to compatible fallback, host-native fallback, or advisor output according to the ladder.
• Use compatible providers and host-native image tools only as fallback executors; Yali inspiration, case search, category matching, live templates, and prompt construction remain first.
• Run through bundled Python or Node CLIs for generation, inspiration search, prompt archive, compatible fallback, and localization.
• Route PPT, slides, deck, and presentation requests to references/ppt-generation/.
• Keep API keys out of repositories and generated examples.

Install Options

Recommended: use the agent-first guide so the current environment can choose between the skills CLI, this package's NPM installer, or a no-NPM GitHub copy:

https://raw.githubusercontent.com/pptt121212/yaliai-gpt-image-2-inspiration/main/docs/install.md

Install through the open skills CLI when npx is available:

npx skills add pptt121212/yaliai-gpt-image-2-inspiration --skill yaliai-gpt-image-2-inspiration --agent claude-code codex --global --yes --copy

Codex only:

npx skills add pptt121212/yaliai-gpt-image-2-inspiration --skill yaliai-gpt-image-2-inspiration --agent codex --global --yes --copy

Yali NPM package installer:

npx @yaliai/gpt-image-2-inspiration install codex

Other NPM installer targets:

npx @yaliai/gpt-image-2-inspiration install all
npx @yaliai/gpt-image-2-inspiration install claude-code
npx @yaliai/gpt-image-2-inspiration install opencode
npx @yaliai/gpt-image-2-inspiration install gemini

No Node/NPM required path:

git clone https://github.com/pptt121212/yaliai-gpt-image-2-inspiration.git
mkdir -p ~/.codex/skills/yaliai-gpt-image-2-inspiration
cp -R yaliai-gpt-image-2-inspiration/SKILL.md \
      yaliai-gpt-image-2-inspiration/agents \
      yaliai-gpt-image-2-inspiration/references \
      yaliai-gpt-image-2-inspiration/scripts \
      ~/.codex/skills/yaliai-gpt-image-2-inspiration/

API Key

Public inspiration search is keyless. Image generation through Yali uses the user's own Yali key:

export YALIAI_API_KEY="your_key_here"

Get your key after login at Yali Skill/API setup.

When asking an AI coding tool to configure the key, tell it to use the variable name YALIAI_API_KEY and follow docs/install.md. The agent should write it to the current user's shell profile, Windows user environment, service environment, or the tool's documented local secrets/runtime environment, then verify that YALIAI_API_KEY is readable. Keep real keys out of repositories, package files, README files, Skill files, and shared prompts.

Compatible fallback execution is optional and does not replace Yali. Enable it only when you want an OpenAI-compatible executor after Yali prompt construction:

export YALIAI_ALLOW_COMPAT_PROVIDER=1
export OPENAI_API_KEY="your_compatible_provider_key"
export OPENAI_BASE_URL="https://api.openai.com/v1"
export OPENAI_IMAGE_MODEL="gpt-image-1"

Image Generation Workflow

flowchart TD
  A["User image task"] --> B["Classify: generate, edit, prompt, batch, or PPT visual"]
  B --> C["Search Yali prompt examples/categories/templates when useful"]
  C --> D["Write production-ready prompt or edit spec"]
  D --> E["Archive prompt/spec"]
  E --> F["Provider ladder preflight: Yali, compatible, host-native, advisor"]
  F --> R{"Recommended executor"}
  R -->|Yali API| G["Yali API runner: queued task + task_id + result URL"]
  G --> H["Localizer: stable local file + Markdown absolute path"]
  R -->|fallback available| I["Compatible or host-native fallback with Yali-built prompt"]
  I --> H
  R -->|setup needed| J["Prompt/spec plus concrete setup guidance"]

Bundled runtime scripts:

| Capability | Python | Node | | --- | --- | --- | | Yali generation, editing, status, result, polling | scripts/python/yali_image_api.py | scripts/node/yali_image_api.mjs | | Inspiration search, categories, case details, templates | scripts/python/yali_inspiration.py | scripts/node/yali_inspiration.mjs | | Result localization to stable files and Markdown previews | scripts/python/localize_image_result.py | scripts/node/localize_image_result.mjs | | Provider ladder inspection | scripts/python/image_provider_ladder.py | scripts/node/image_provider_ladder.mjs | | Prompt/spec archive | scripts/python/archive_prompt.py | scripts/node/archive_prompt.mjs | | Compatible fallback execution | scripts/python/compatible_image_api.py | scripts/node/compatible_image_api.mjs | | Install/runtime regression check | - | scripts/node/self_test.mjs |

Run commands from the Skill directory. Prefer Python when python3 exists; use Node when Python is unavailable. If neither runtime exists, the Skill can still return prompts and setup guidance but cannot execute local image generation.

Before generation, pass the current host image-tool state to the provider ladder with --host-native-available or --host-native-unavailable. The ladder reports checked_providers[], yali_api_status, compatible_fallback_available, host_native_status, provider_order, recommended_mode, and recommended_reason. Detailed CLI parameters are defined in SKILL.md and references/image-generation-workflow.md.

Real Image Examples

| Use case | Example | | --- | --- | | Product and infographic | Bed set price infographic | | UI mockup | YouTube homepage mockup | | Visual guide / infographic | Image generation editing guide | | Storyboard | MV piano scene storyboard | | Advertising / product | Classical temple perfume ad | | Portrait photography | Summer beach portrait |

More examples: docs/examples.md

PPT Generation Branch

When the user asks for PPT, slides, deck, keynote, or presentation output, the main Skill routes the agent to references/ppt-generation/ for a local PPT workflow.

The PPT branch supports a local workflow:

1. Plan the deck as slides_plan.md and slides_plan.json.
2. Generate one 16:9 image prompt per slide.
3. Use the Yali API first for slide images; when explicitly allowed, use compatible fallback execution with the same Yali-built prompts.
4. Package the images into index.html and an image-based presentation.pptx.

PPT examples: docs/ppt-examples.md

Yali Free Image API

Base URL:

https://www.yaliai.com/wp-json/yali/v1

Useful public endpoints:

GET /inspiration/categories
GET /inspiration/search?q=poster&limit=10
GET /inspiration/cases/{case_id}
GET /free-image/api/templates
GET /free-image/api-docs

Generation endpoints require:

Authorization: Bearer $YALIAI_API_KEY

Public Library Coverage

The public inspiration library currently covers categories such as:

• Portraits and photography
• Illustration and art styles
• Product, e-commerce, and packaging
• Image editing and reference-image control
• Posters, covers, and advertising
• Natural landscapes and locations
• Social media, live-stream, and screenshot styles
• Architecture, interior, and spatial design
• Infographics and structural diagrams
• Documents, receipts, and handwriting
• Typography and text effects
• Storyboards and motion reference
• Branding and visual identity
• Product UI and interaction design

Best For

• Image generation through the bundled Yali API workflow
• Image editing, retouching, object/background replacement, and reference-image tasks
• Image prompt research, writing, improvement, and comparison
• Product shots and e-commerce visuals
• Posters, banners, covers, ads, and social media graphics
• UI mockups and interface concepts
• Infographics, diagrams, and technical explainers
• Logo and brand direction exploration
• Storyboards and scene planning
• Yali API image generation with localized Markdown previews
• Local PPT generation workflows powered by slide images

Package Contents

SKILL.md
agents/openai.yaml
scripts/python/localize_image_result.py
scripts/python/yali_image_api.py
scripts/python/yali_inspiration.py
scripts/node/localize_image_result.mjs
scripts/node/yali_image_api.mjs
scripts/node/yali_inspiration.mjs
scripts/node/self_test.mjs
references/api.md
references/prompt-workflow.md
references/image-generation-workflow.md
references/ppt-generation/*.md
docs/*.md
bin/install.mjs

License

MIT