skills.rest
tribgames
tribgamesOFFICIAL
Explore all skills by this author →
GitHub

image

Generate and edit still images through the media tool with structured prompts and reference controls.

6|Updated Jun 25, 2026
Search Tags:#prompt engineering#image generation#provenance#image editing#reference images#aspect ratio#media tool
One-click install
npx skills add https://github.com/tribgames/mixdog --skill image-tribgames
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: image
Source: https://github.com/tribgames/mixdog/tree/main/src/defaults/skills/image
Command: npx skills add https://github.com/tribgames/mixdog --skill image-tribgames

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Producing usable AI-generated images requires disciplined prompt structure, correct aspect and resolution choices, and verification of results; this Skill standardizes that entire workflow so generated or edited images actually match the brief and land correctly in the deliverable. ## Core Features & Use Cases - Structured prompt form: Labeled lines (Use, Scene, Subject, Style, Composition, Text, Keep, Avoid) in priority order, with anti-pattern rules that replace keyword soup and quality tokens with narrative descriptions. - Edit briefs with locks: Edits name the exact change plus a preserve list (face structure, logo geometry, palette, lighting) repeated on every retry, with one corrected retry per job. - Reference image handling: Each reference gets an index and role (base scene, subject identity, style only), capped by the model's maxReferences, with identity-first ordering. - Placement-driven output rules: Aspect ratio comes from the destination frame (16:9 hero, 9:16 story, 1:1 icon, 1.91:1 OG card), exact copy is composited rather than generated, and results are verified by reading the file before use. - Use Case: A landing page needs a hero image. The Skill lists signed-in image lanes, generates a 16:9 prompt-structured image with no text or logos, inspects the result, and saves it beside the page with lane/model provenance recorded. ## Quick Start Use the image skill to generate a 16:9 hero image of a ceramic coffee mug on a wooden table in warm morning light, saved as hero.png.

Frequently Asked Questions about image

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

FAQPage Schema
How do I generate an image with the media tool?▼

Call media with action 'list' and kind 'image' once per session to see signed-in lanes and the remembered model, then call action 'generate' with a structured prompt, output path, and controls from that model's list. Lane and model are omitted by default so the user's Studio selection applies.

How do I edit an existing image without changing everything?▼

Write an edit brief naming the desired result, the single change, and a preserve list of locks such as face structure, pose, palette, and lighting. Pass the source image in references and repeat the same lock list on any retry.

What aspect ratio should I use for generated images?▼

Aspect comes from where the image will live: 16:9 or 3:2 for heroes, 9:16 for mobile stories, 1:1 for icons and avatars, and the closest listed ratio to 1.91:1 for OG cards. Never reuse a ratio from a previous generation.

Can AI image generation render exact text like packaging copy?▼

No. Copy that must be exact, such as packaging, UI, or legal text, is generated as an empty reserved area and composited afterwards with a real font. Generated text is only used for short decorative strings and is always inspected for garbled glyphs.

Why does requesting a transparent background fail?▼

Image models do not produce real alpha channels and instead burn a fake checkerboard into the picture. Generate on a stated solid background such as pure white, then remove the background afterwards with an image tool.

What happens when no image lane is signed in?▼

When list or a failed generate reports no image lane, the workflow continues without a generated picture and says so. It never substitutes a placeholder, stock-photo URL, or emoji.

Related Skills

Explore More Design & Creative Skills→
forcedotcomforcedotcom
803

searching-media

Search Salesforce CMS and Data 360 for existing visual media assets.

Official
Intermediate
JustinChangTWJustinChangTW

video-presentation

Convert slides and narration into timed presentation videos with captions and MP4 export.

Community
Intermediate
cosmicstack-labscosmicstack-labs
364

hyperframes-media

Generate local TTS audio, word-timestamp transcripts, and transparent overlays for media compositions.

Official
Intermediate
Mathews-TomMathews-Tom
310

concept-to-image

Convert concept descriptions into self-contained HTML/CSS/SVG visuals and export as PNG or SVG files.

Community
Advanced