Maintenance rules

Skill motiful/readme-craft/.claude/skills/maintenance-rules

Layout-first README generation for AI agents — organize by how readers scan, not by what you want to say. 3-tier layout strategy with 45-point quality audit.

Install

npx -y skills add motiful/readme-craft --skill maintenance-rules

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Maintenance rules for readme-craft. MUST keep quality checklist item count consistent across SKILL.md, README, and case study. MUST regenerate docs/ images when referenced content changes. MUST keep templates aligned with current methodology. MUST keep SKILL.md under 500 lines. Triggers on "update readme-craft", "maintain readme-craft", "check readme-craft consistency".

SKILL.md

3.6 KB, as published. Nobody here has run it

readme-craft Maintenance Rules

Constraints and procedures for maintaining the readme-craft repository.

Execution Procedure

maintain(trigger) → updated | no_change

if change proposed:
    run Decision Test (constraints below) → accept | reject
    verify all Constraints (MUST/NEVER)
    run Consistency Checks
    update Changelog (max 5, trim oldest)

Constraints

MUST NOT push SKILL.md body over 500 lines (currently 487)
MUST keep quality checklist item count consistent across: SKILL.md body, references/quality-checklist.md, README features, docs/case-study-skill-forge.md scoring tables. When checklist changes, update all four
MUST regenerate docs/ images when the content they illustrate changes. Specifically: docs/skill-forge-comparison.png must match docs/case-study-skill-forge.md — regenerate via node scripts/generate-comparison.mjs when the case study's Before or After source changes
MUST use the After markdown from its original repo location when generating comparison screenshots — resolveLocalPaths resolves src/srcset relative to the markdown file's directory. Copying to /tmp/ breaks image resolution (e.g., .github/logo-dark.svg becomes /tmp/.github/logo-dark.svg). Correct: --after /path/to/original/repo/README.md
MUST keep templates (assets/universal-readme.md, assets/skill-readme.md) aligned with current methodology. When SKILL.md rules change (e.g., section ordering, <details> policy, Checkout Test criteria), update templates to match
MUST keep SKILL.md ↔ README ↔ references terminology consistent
MUST add References table entry when adding new reference files
MUST verify reference files with >100 lines have a TOC
NEVER let docs/ case studies praise or recommend patterns that SKILL.md forbids

Consistency Checks

Quality checklist dimensions and item counts match across all references
SKILL.md EP step numbers are sequential and complete
All SKILL.md reference file pointers resolve to existing files
docs/logo-gallery.md references SVGs that exist in examples/logos/
examples/logos/manifest.json uses relative paths, not absolute
.github/repo-meta.yml description aligns with README one-liner

Update Triggers

Event	What to check
Quality checklist changes (add/remove/reorder items)	SKILL.md body, README features, case study scoring tables, references/quality-checklist.md
Methodology change (layout rules, formatting policy)	Templates in `assets/`, case study, docs/how-it-works.md
Logo engine changes (presets, palettes, CLI flags)	references/logo-generation.md, references/logo-examples.md, docs/logo-gallery.md, package.json scripts
Case study Before or After source changes	Regenerate docs/skill-forge-comparison.png
New reference file added	SKILL.md References table, TOC if >100 lines
Badge ecosystem changes	references/badges.md

Changelog (max 5 entries)

2026-03-26: Initial creation. Triggered by self-review finding: case study praised <details> (forbidden by SKILL.md), comparison image showed stale methodology, quality checklist count inconsistency across files. All three caused by missing maintenance-rules enforcement.