Iterative Design Assistant

Track design iteration history and apply consistent styling decisions across revision cycles. Use when handling follow-up change requests that reference previous decisions (“前回も色で良いんだけど”, “same style as before”).

No API Required

Download Skill Package (.skill) View Source on GitHub

Table of Contents

1. Overview

Maintains a session-local design decision log (JSON v1.0) across 5 categories (color/typography/layout/content/style). CLI commands cover init/record/query/search/apply/history/token/resolve. Resolves contextual references in JP and EN, manages design tokens by category namespace, and provides bidirectional traceability between decisions and elements.

2. Prerequisites

  • Python 3.9+
  • No API keys required

3. Quick Start 

# Install the skill locally
make install SKILL=iterative-design-assistant

# Or fetch the .skill package
curl -L -o iterative-design-assistant.skill https://github.com/takusaotome/claude-skills-library/raw/main/skill-packages/iterative-design-assistant.skill

Then trigger the skill in Claude Code by describing what you want — see the Usage Examples section below for trigger phrases.

4. How It Works

The skill follows the workflow documented in its SKILL.md. Key stages:

  1. Input parsing — interprets the user request and any provided source files.
  2. Core processing — applies the skill’s domain logic (see Reference section).
  3. Output generation — produces structured artifacts (markdown / JSON / templates) ready for downstream use.

For the authoritative step-by-step procedure, open skills/iterative-design-assistant/SKILL.md.

5. Usage Examples

  • A reviewer says “same as last time” and you need to recall what “last time” was
  • You’re iterating on slide / brand / layout designs across multiple sessions
  • You want a queryable history of design decisions per project
  • You need design tokens consistent across an iteration

6. Understanding the Output

The skill produces structured output following the conventions in its templates and reference docs (see Section 10). Outputs are:

  • Reproducible — identical input + same templates → same output structure.
  • Reviewable — each section is labeled and ordered consistently.
  • Composable — outputs of this skill can feed adjacent skills (see Section 8).

7. Tips & Best Practices

  • Start with a small, realistic input to validate the workflow before scaling.
  • Keep skills/iterative-design-assistant/SKILL.md open alongside this guide; it remains the authoritative source.
  • Read the most relevant reference file first (see Section 10) instead of trying to absorb all of them.
  • Run scripts on test data before applying to production-bound inputs.
  • Preserve intermediate outputs so you can explain assumptions and trace decisions.

8. Combining with Other Skills

  • Pair with adjacent skills in the same category to cover the planning → execution → review arc.
  • Browse the Meta & Quality category for neighboring workflows: category index.
  • See the full English skill catalog: skill catalog.

9. Troubleshooting

  • Re-check prerequisites first; missing runtime dependencies are the most common failure mode.
  • Run helper scripts on a minimal input before applying them to a full dataset.
  • Compare your input shape against the reference files to confirm expected fields, sections, or metadata.
  • Confirm Python version (3.9+) and required packages are installed in the active environment.
  • When output looks incomplete, re-read the relevant reference file to verify the input contract.

10. Reference

References:

  • skills/iterative-design-assistant/references/design-decision-methodology.md

Scripts:

  • skills/iterative-design-assistant/scripts/design_log.py

Assets:

(none)