AI Skill Report Card

Building Digital Color Systems

A-80·Nov 20, 2026·Source: Web
14 / 15

A user wants to sell "a skin tone palette." Don't ship 20 hex codes — ship a system:

  1. Pick a product line (Illustration/SSS, Print-Safe/Prepress, or UI/UX Tokens)
  2. Feed source material into Gemini Notebook with the matching extraction prompt (below)
  3. Run the Python build script on the extracted JSON to generate .swatches + metadata
  4. Bundle into the 4-folder ZIP structure with docs and brush presets
  5. Price at $19–$49 using the problem/system/contents/specs listing structure
Recommendation▾
Add a third example showing a failure/bad outcome (e.g., what happens when TAC validation is skipped) to strengthen the examples section with contrast, not just success cases.
14 / 15

Progress:

  • Step 1: Select target specialty (Illustration, Print, or UI/Planner)
  • Step 2: Gather 3–5 dense source documents (rendering guides, color science papers, print specs, accessibility standards)
  • Step 3: Run the matching extraction prompt in Gemini Notebook/NotebookLM
  • Step 4: Validate output against the constraint rules (hue-shift direction, TAC limits, WCAG ratios)
  • Step 5: Run build_asset_bundle() to generate .swatches + metadata JSON
  • Step 6: Assemble the 4-folder ZIP (Palettes / Brushes / Documentation / Automation)
  • Step 7: Write listing copy using the 4-section structure
  • Step 8: Set price based on defensive moat strength (see table below)

Step 1: Choose the Product Line

SpecialtyDeliverablePriceMoat
Character/SSS RenderingSwatches + Brush Pack + PDF$19–$29Rendering logic tied to brush dynamics
Print/PrepressCMYK gamut files + Vector swatches + Spec sheet$24–$39Technical compliance (zero reprint errors)
UI/Brand TokensJSON tokens + Goodnotes/Notion kit + WCAG report$34–$49+Turnkey + accessibility verification

Never ship a bare .swatches file alone — that's the $3–$5 commodity tier. The moat is always the system layer (brush pairing, print compliance, or accessibility proof) wrapped around the raw hex data.

Step 2–3: Extraction Prompts

For Illustration/SSS systems, use Prompt 1 — extracts 4 tiers (Base Local, SSS Transition, Ambient Shadow, Specular Rim), each with Hex/HSB/CMYK plus matched Procreate Brush Studio settings (Wet Mix, Dynamics, Rendering Mode). Hard constraint: shadows must cool-shift, SSS zones must warm-shift relative to base.

For Print/Prepress systems, use Prompt 2 — extracts a master table (Swatch Name, Hex, CMYK, TAC %, Gamut Pass/Fail) grouped into Primary Inks / Accent Overprints / Background Stock, plus halftone screen angles. Hard constraint: reject any swatch with TAC > 280% or outside CMYK gamut.

For UI/Token systems, adapt the same pattern: extract paired background/text tokens with WCAG 2.1 contrast ratios, dark/light mode pairs, and accessibility pass/fail flags instead of brush settings.

Always specify the target complexion/medium/print method in brackets before running — generic prompts produce generic (uncommercial) output.

Step 4: Validate Before Packaging

Reject extraction output that violates the prompt's constraint clause (wrong hue-shift direction, TAC overage, WCAG failure). Re-run the prompt with corrected source excerpts rather than manually patching hex values — manual patches break the internal logic that justifies premium pricing.

Step 5: Build the Bundle

Convert validated JSON color data into deliverables using hex_to_hsb() + build_asset_bundle():

Python
sample_colors = [ {"name": "Base Light", "hex": "#F2D6C1", "role": "Base Local"}, {"name": "SSS Transition", "hex": "#E08D79", "role": "Subsurface Scattering"}, {"name": "Cool Shadow", "hex": "#6B5B6E", "role": "Ambient Shadow"}, {"name": "Rim Highlight", "hex": "#FFF5E4", "role": "Specular Highlight"} ] build_asset_bundle("Character_Rendering_System", sample_colors, "./output_bundle")

This produces {bundle_name}.swatches (native Procreate payload) and {bundle_name}_metadata.json (role-tagged sidecar for cross-platform conversion to .aco/.ase).

Step 6: Package the ZIP

Product_Bundle/
├── 01_Palettes/           (.swatches, .aco, .ase)
├── 02_Brushes_and_Presets/ (.brushset, .tpl)
├── 03_Documentation/      (User guide PDF, JSON metadata, compliance report PDF)
└── 04_Automation_Scripts/ (generate_swatches.py)

Documentation folder is non-optional — it's the primary justification for price above the $5 commodity ceiling.

Step 7: Write Listing Copy

Title formula: [System Name] + [Target Application] + [Problem Solved]

Four-section body: Problem (specific time/quality pain) → System Solution (the engineering behind it) → What's Inside (itemized file list with counts) → Technical Specs (platform compatibility, color space, gamut safety).

Recommendation▾
Include the actual `hex_to_hsb()` and `build_asset_bundle()` function implementations or at least signatures/docstrings, since the skill references them as core tooling but only shows a call site.
14 / 20

Example 1: Input: "I have a watercolor pigment research PDF and want to sell a palette for gouache illustrators." Output: Product Line A variant — run Prompt 1 with [SPECIFY COMPLEXION/MEDIUM] set to "gouache pigment granulation," extract 4-tier system with brush-equivalent "Wet Mix/Bleed" mapped to gouache's granulation/wet-edge behavior, package as $19–$29 bundle titled "Gouache Granulation Color System | Pigment Ramps & Blend Presets."

Example 2: Input: "I want to make a risograph color pack from an old CMYK ink catalog." Output: Run Prompt 2 with target "RISOGRAPH," extract Primary Inks/Accent Overprints/Background Stock table, enforce TAC ≤ 280%, include screen angle recommendations, price $24–$39, title "Risograph-Safe Ink Matrix | Zero-Reprint CMYK Swatch System."

Recommendation▾
Tighten the description slightly — it's dense and covers three distinct use-cases (build/price/package, extraction via NotebookLM, upselling from swatches) which could be split into clearer trigger phrases.
  • Always pair raw color data with a behavioral layer (brush settings, print specs, or accessibility proof) — this is the entire defensive moat.
  • Enforce the prompt's built-in constraints programmatically where possible (hue-shift direction, TAC ceiling, contrast ratio) rather than trusting manual review.
  • Ship metadata JSON alongside .swatches always — it enables cross-platform repackaging (Photoshop/Illustrator/CSP) without re-extraction, which increases perceived value.
  • Match price to moat strength: brush-dependent systems (easier to approximate) sit at the lower end ($19–$29); compliance-verified systems (harder to fake) command $34–$49+.
  • Keep source documents dense and technical (rendering papers, prepress specs, WCAG docs) — thin source material produces thin, uncommercial extraction output.
  • Selling the .swatches file alone with no brush/compliance/accessibility layer — this is the $3–$5 commodity trap the whole strategy exists to avoid.
  • Skipping the constraint-validation step and shipping extraction output that technically violates print gamut or WCAG contrast — destroys the "zero error" value proposition.
  • Omitting the metadata JSON sidecar, which locks the product to one platform and kills cross-tool resale value.
  • Generic extraction prompts without a specified target medium/complexion/print method — produces generic, easily-replicated output that can't justify premium pricing.
  • Listing copy that leads with feature lists instead of the specific workflow problem — buyers pay for time/error savings, not hex counts.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
14/15
Workflow
14/15
Examples
14/20
Completeness
17/20
Format
14/15
Conciseness
13/15