reverse-engineer

Extract a supplied image into structured JSON with typography, palette, grading, composition, named elements, inferred layers, lighting, and evidence confidence.

Jump to install Review files

Repository: codejunkie99/image-loop
Last source activity: September 9, 2026 at 11:57
Detected SKILL.md language: English
Stars: 13
Forks: 6

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Install with an AI assistant

Copy this prompt into the AI assistant you're using.

Follow https://skillsmp.com/skill-install/prompt.md to install only the Agent Skill at https://github.com/codejunkie99/image-loop/tree/main/skills/reverse-engineer.

A direct command skips the review prompt. Inspect the source before running it.

npx skills add https://github.com/codejunkie99/image-loop --skill reverse-engineer

The command stays on one line. Scroll horizontally to inspect it before copying.

Prefer a local copy? Download the files currently available to SkillsMP.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

SKILL.md

Source instructions · Read-only preview

name: reverse-engineer
description: Extract a supplied image into structured JSON with typography, palette, grading, composition, named elements, inferred layers, lighting, and evidence confidence.

# Reverse Engineer an Image Inspect the attached image and follow the companion [Image Edit Map reverse-engineering protocol](../image-edit-map/references/reverse-engineer.md). Use its [JSON Schema](../image-edit-map/references/image-spec.schema.json). The installer includes that companion; if it is missing, say which file is needed rather than inventing the schema. A bare invocation requests a comprehensive visual breakdown. Ask for an image if none is available. Return JSON plus a short readable explanation unless the person requests JSON only. Unknown values are null and explicitly marked unknown. Do not claim to recover the original prompt, exact fonts, hidden content, actual editable layers, or exact grading parameters from a flattened picture. Offer the numbered map or an edit only after satisfying the extraction request. Do not generate or modify artwork during a JSON-only analysis. Claude Code: `/reverse-engineer`. Codex: `$reverse-engineer` or select the skill through `/skills` where available.

View on GitHub

Back to search Back to installation

image-reconstruction-skill

Map reference images into selectable named elements, produce evidence-labeled reconstruction JSON, and compile it into prompts for reconstruction, remixing, and scoped edits. Use when a person wants to understand, recreate, or selectively change an image.

2026-09-10

image-reconstruction

Analyze image references into named visual elements and reconstruction JSON, then compile prompts for recreating, remixing, or selectively editing them. Use when the user wants to understand a reference, reuse its visual decisions, or preserve specific parts during an edit.

2026-09-10

image-loop

Generate or edit images with a bounded create-review-repair loop, an independent vision reviewer, explicit preservation checks, and saved evidence. Use for image work that needs iterative checking, or invoke image-loop with a brief or reference image.

2026-09-09

View all from this repository

reverse-engineer

Source facts

Install options

Install with an AI assistant

Review the source files

More from this repository

Install with an AI assistant

More from this repository