feat: add & fortify color-management skill — ICC profiles, working spaces, gamut analysis #67

Merged

magnus merged 2 commits from feat/color-management-skill into main

2026-07-05 16:09:30 -04:00

jasper commented

2026-07-05 16:06:20 -04:00

Contributor

Summary

Creates the color-management skill for the agent-skills repo — expert-level ICC profile color management for open-source workflows. Followed by a 10-phase fortification filling all identified gaps.

What's included (20 files, 3.6K+ lines)

Core SKILL.md

Quick reference table (15 common tasks with their reference/script)
Gotchas section (sRGB variants, matrix perceptual lie, unbounded limits, etc.)
3 concrete failure examples with explanations

7 Python Scripts

icc-profile-inspect.py — extract ICC metadata, primaries, TRC
well-behaved-check.py — xicclu neutral axis verification
color-space-convert.py — profile-to-profile with intent/BPC/unbounded
gamut-check.py — out-of-gamut pixel analysis + overlays
srgb-compare.py — sRGB variant comparison
color-difference.py — CIE76/CIE94/CIEDE2000/CMC l:c
color-report.py — unified report combining all tools

9 Reference Files

color-management-overview, icc-profile-operations, tool-reference
working-spaces-reference (with decision tree), glossary
soft-proofing-workflow, monitor-calibration-workflow
dcraw-pipeline, hex-quantization-and-profile-creation

2 Templates

icc-profile-report.md, lch-layer-stack.md

Key Sources

Distilled from Elle Stone's ninedegreesbelow.com and Bruce Lindbloom's color science resources
All scripts check for required tools with install hints
No personal/homelab info — fully public-consumable under MIT

## Summary Creates the `color-management` skill for the agent-skills repo — expert-level ICC profile color management for open-source workflows. Followed by a 10-phase fortification filling all identified gaps. ## What's included (20 files, 3.6K+ lines) ### Core SKILL.md - Quick reference table (15 common tasks with their reference/script) - Gotchas section (sRGB variants, matrix perceptual lie, unbounded limits, etc.) - 3 concrete failure examples with explanations ### 7 Python Scripts - `icc-profile-inspect.py` — extract ICC metadata, primaries, TRC - `well-behaved-check.py` — xicclu neutral axis verification - `color-space-convert.py` — profile-to-profile with intent/BPC/unbounded - `gamut-check.py` — out-of-gamut pixel analysis + overlays - `srgb-compare.py` — sRGB variant comparison - `color-difference.py` — CIE76/CIE94/CIEDE2000/CMC l:c - `color-report.py` — unified report combining all tools ### 9 Reference Files - color-management-overview, icc-profile-operations, tool-reference - working-spaces-reference (with decision tree), glossary - soft-proofing-workflow, monitor-calibration-workflow - dcraw-pipeline, hex-quantization-and-profile-creation ### 2 Templates - icc-profile-report.md, lch-layer-stack.md ### Key Sources - Distilled from Elle Stone's ninedegreesbelow.com and Bruce Lindbloom's color science resources - All scripts check for required tools with install hints - No personal/homelab info — fully public-consumable under MIT

jasper added 2 commits

2026-07-05 16:06:20 -04:00

feat: add color-management skill (squash with fortification) c05bf92312

feat: fortify color-management skill — 10 phases abfd1dd5e1

Phase 1 — CIEDE2000/CIE94/CMC in color-difference.py with colour-science
  fallback and pure Python implementations
Phase 2 — Soft proofing workflow reference (NULL Curves, 4 strategies,
  LCMS2 linear gamma bug, gamut check automation)
Phase 3 — Monitor calibration workflow (ArgyllCMS dispcal/dispread/colprof,
  DisplayCAL GUI guide, colorimeter correction matrices)
Phase 4 — dcraw pipeline reference (raw → linear TIFF → profile assign →
  working space convert, negative tristimulus problem)
Phase 5 — Hex quantization reference (prequantized primaries methodology,
  which profiles affected, ArgyllCMS prequantization)
Phase 6 — GIMP LCH layer stack template (3-group architecture, Chroma mask,
  Luminance desaturation, out-of-gamut monitoring)
Phase 7 — Working space decision tree (trigger-based flow, quick reference
  table, why-choose guides for common spaces)
Phase 8 — Concrete failure gotchas (gamma slider + unbounded sRGB = magenta,
  wrong-working-space color correction, channel-mixing out of gamut)
Phase 9 — Unified color-report.py (combines inspect + well-behaved +
  gamut check into single report)
Phase R — DisplayCAL research incorporated into monitor calibration reference

magnus merged commit 13801a38a4 into main

2026-07-05 16:09:30 -04:00

magnus referenced this pull request from a commit

2026-07-05 16:09:31 -04:00

Merge pull request 'feat: add & fortify color-management skill — ICC profiles, working spaces, gamut analysis' (#67) from feat/color-management-skill into main