beautify-github-readme
An agent skill (oil-oil, MIT) that redesigns a repository’s README with project-native visual design and a clearer information hierarchy — the spoke’s first skill aimed at repo/README presentation as its vertical. Not a library you import; a skill an agent invokes.
What it does
Two modes: Whole-README restructures content order and hierarchy (moving proof and value propositions ahead of technical details) and adds a full visual system; SVG-only generates decorative assets (heroes, section headers, workflow diagrams) without touching the existing copy. Its guiding principle is a separation of concerns: Markdown carries the searchable body text and commands, SVG carries visual identity and complex diagrams — so the result stays maintainable and searchable rather than an image-only banner. Python- based; emits GitHub-safe SVG/PNG/WebP + Markdown, and derives each project’s visual language from its own identity rather than a generic template.
Where it fits
It extends the spoke’s skills-generalize-to-verticals thread past coding/PM/design into developer communication / repo presentation — a sibling of designpowers (design), renwei-writing (writing), ai-job-search (job apps), and pm-skills (PM). Its “proof and value before technical detail” reordering rhymes with renwei-writing‘s editing-with-judgment and ponytail‘s restraint: a skill encoding an editorial discipline, not just generation. Two lighter adjacencies: it’s a docs-as-artifact cousin of openwiki / autowiki-factory (machine-generated repo documentation) but aimed at the human-facing README rather than agent-grounding docs; and the Markdown-for-text / SVG-for-identity split is a small signal-vs-noise-adjacent idea applied to a page instead of a context window. T2: a primary MIT repo with documented behavior, self-described and unbenchmarked.
Related
agent-skills · agentskills-spec · oil-oil · designpowers · renwei-writing · ai-job-search · openwiki