jimliu/baoyu-skills
21 skills383k installs439k starsGitHub
Install
npx skills add https://github.com/jimliu/baoyu-skillsSkills in this repo
1Baoyu Image Genbaoyu-image-gen is an agent skill package for solo builders who need production-style AI image generation without juggling ten different API docs in chat. It documents how to call major hosted image models through official APIs, including Azure and regional providers, and supports text-to-image, reference-guided generation, aspect ratio control, and reading prompts from files for repeatable batches. Sequential generation is the default; when you already have multiple prompts or want steadier throughput, parallel batch mode is the intended path. The skill also standardizes how the agent should ask clarifying questions—preferring runtime user-input tools with a numbered fallback. Prism lists it for discovery, but upstream metadata explicitly deprecates this skill in favor of baoyu-imagine, so treat this entry as a bridge for existing installs and migrate when you start new stacks.24.4kinstalls2Baoyu Markdown To HtmlBaoyu-markdown-to-html is a Baoyu-skills converter that transforms Markdown sources into self-contained, styled HTML aimed at WeChat Official Account constraints and other platforms that reject external stylesheets. Solo builders writing launch posts, changelog pages, or course modules can keep authoring in Markdown while shipping presentation-ready HTML with syntax highlighting, math, diagrams, and footnotes. The skill documents when users ask for markdown to html, md 转 html, or WeChat-style bottom citations for outbound links. Agents resolve script paths relative to SKILL.md and prefer runtime ask-user tools for theme and option selection. It complements design-heavy cover generation elsewhere in the same repo but focuses on faithful typography and embed-friendly markup. Expect bun or npx on the machine; output is a finished HTML artifact you paste into CMS, email, or公众号 backends rather than a live dev server.24.3kinstalls3Baoyu Cover ImageBaoyu-cover-image is a Baoyu-skills generator that produces article cover images from structured briefs across five dimensions—content type, palette, rendering style, on-image text, and mood—so solo builders can ship credible visuals for blogs, newsletters, and social posts. It exposes eleven palettes and seven rendering styles with cinematic, widescreen, or square crops suited to video thumbnails, OG images, and square feeds. Invoke it when users say generate cover image, create article cover, or make cover. The skill prioritizes runtime ask-user tools to capture creative choices in one batched interaction, then routes to the configured image backend. It pairs naturally after markdown-to-html in the same monorepo when you are packaging a full publish bundle. This is generative media workflow knowledge, not a stock photo API catalog; you still review brand fit and text legibility before shipping.24kinstalls4Baoyu Article Illustratorbaoyu-article-illustrator is an agent skill that turns your article or outline into a cartoon-style infographic following fixed image specs: landscape 16:9, hand-drawn illustration throughout, and language-matched on-image text. Solo builders and indie marketers use it when publishing tutorials, launch posts, or explainers and need a consistent visual that highlights core concepts without stock photos or generic AI realism. The skill encodes core principles—sparse accents, readable hierarchy, and mono-ink styling—so outputs stay on-brand for technical writing. Invoke when you already have draft copy and want a single shareable hero or inline diagram for social, newsletter, or docs pages.23.8kinstalls5Baoyu Xhs Imagesbaoyu-xhs-images is a narrow agent skill in the baoyu-skills family for producing Xiaohongshu (小红书) visual assets when your runtime lacks a native imagegen tool but has Codex CLI logged in. It tells the agent to prefer invoking baoyu-image-gen with provider codex-cli so retries, caching, batch runs, and EXTEND.md preferences stay consistent with other image providers. If that sibling skill is not installed, the agent may spawn the baoyu-codex-imagegen wrapper directly, honoring BAOYU_CODEX_IMAGEGEN_BIN or discovering packages/baoyu-codex-imagegen under the plugin root. Solo builders shipping content-heavy apps or growth experiments use it to turn structured prompt markdown files into correctly aspect-ratioed slides without hand-rolling shell scripts each session. It assumes Bun for TypeScript entrypoints and absolute paths for outputs and references.23.5kinstalls6Baoyu Slide Deckbaoyu-slide-deck is an agent skill that runs a presentation analysis framework before you create slide decks. Solo and indie builders use it when a pitch, launch talk, webinar, or stakeholder update needs a sharp core message instead of bullet soup copied from docs. The skill walks through message hierarchy—one ≤15-word takeaway, up to three to five supporting points, and a clear call-to-action—then fills an audience decision matrix covering beliefs, desired decisions, barriers, and convincing evidence. An audience adaptation table steers executives toward outcomes and ROI, technical readers toward architecture and specs, and general audiences toward benefits and stories. Install it when you are about to brief an agent or yourself on deck structure for validation conversations, ship-week announcements, or growth content, and you want narrative discipline before opening Keynote, Google Slides, or a design tool.23.3kinstalls7Baoyu Url To Markdownbaoyu-url-to-markdown is an agent skill for solo and indie builders who need dependable webpage-to-markdown capture without hand-copying from the browser. It drives the baoyu-fetch command-line tool, which uses Chrome CDP plus adapters tuned for X/Twitter, YouTube transcripts, Hacker News threads, and a Defuddle path for generic sites. When a page needs login or CAPTCHA, the skill’s interaction wait modes let you complete the gate and still finish extraction. Install bun, wire the vendored CLI under the skill’s scripts, and invoke whenever someone says they want to save a URL as markdown for research, docs, or repurposed content. Compared with one-off fetch snippets, this packages site-specific behavior and user prompting rules so agents batch questions correctly and pick sensible output filenames. It is integration-heavy—not a hosted MCP server—so you review network and browser permissions before running on sensitive accounts.22.7kinstalls8Baoyu ComicBaoyu Comic is an agent skill that implements a deep content-analysis framework for solo builders and creators who want to turn articles, docs, or stories into comics without skipping strategy. Before any art or copy generation, it walks through what the source actually says, who should read it, and which ideas deserve panels versus captions. The workflow spans understanding the central message and key concepts, mapping structure to a narrative arc, harvesting examples that visualize well, and grounding everything in context—authorship, culture, and hidden assumptions. It is meant to be invoked when you have raw material but no clear comic thesis, when audience or tone is fuzzy, or when you need a repeatable brief your coding agent can hand to illustration or scripting steps later. Framed for Prism’s journey, it most often appears while validating and scoping a content product, and remains useful when refining launch and growth content because the same analysis reframes messaging for distribution.22.2kinstalls9Baoyu Post To Xbaoyu-post-to-x is an agent skill that turns a local Markdown article into HTML and walks you through pasting it into X’s Articles editor with images and rich formatting. It is aimed at solo builders and indie marketers who already write in Markdown and want to ship the same piece on X without rebuilding layout by hand. The workflow is deliberately browser-centric: you pick one of three control paths—Codex Chrome Plugin (when you ask for @chrome or the Codex Chrome extension), Chrome Computer Use, or a CDP script path—and the skill refuses to use the generic in-app browser for this job. Supporting scripts convert the file, persist HTML under /tmp, and copy HTML to the clipboard for a single paste into X. Use it when a draft is ready and distribution on X Articles is the next step; it does not replace SEO strategy, scheduling, or thread-style short posts.21.8kinstalls10Baoyu Danger X To Markdownbaoyu-danger-x-to-markdown is an agent skill package centered on a mandatory first-time setup for X-to-markdown conversion preferences. Before any URL conversion, media download, or output path work, the agent must complete an AskUserQuestion flow and write EXTEND.md—no parallel conversion steps are allowed. Solo builders use it when they want repeatable rules for whether images and videos land in local imgs/ and videos/ folders or stay as remote URLs, and where markdown files default on disk. The setup is intentionally blocking so later runs stay consistent without re-prompting every session. It pairs well with content repurposing pipelines: scrape or paste X links, normalize to markdown, then feed blogs, newsletters, or knowledge bases. The skill is workflow-oriented rather than a hosted API; you bring URLs and a writable project directory. After EXTEND.md exists, the conversion workflow continues with your saved language and media choices.21.2kinstalls11Baoyu Compress Imagebaoyu-compress-image is a practical compression skill backed by a Bun script that solo builders can run in a repo or CI prep step to cut image weight before ship. It detects the best available compressor on your machine—sips on Darwin, cwebp for WebP, ImageMagick convert, or Sharp as fallback—and supports common raster extensions with quality and format flags. Recursive directory runs and JSON output make it easy to log savings across a marketing site or app screenshot folder without paying for a hosted optimizer. It fits the Build phase when you are tightening Core Web Vitals, prepping store assets, or batch-converting screenshots for docs. Expect shell and filesystem access because the tool spawns local CLI utilities.21.1kinstalls12Baoyu Danger Gemini Webbaoyu-danger-gemini-web is an agent skill package that connects your Claude Code, Cursor, or Codex workflow to Google Gemini’s web image generation surface through Chrome DevTools Protocol automation. Solo builders use it when they need on-demand visuals for landing pages, social posts, or in-app placeholders without standing up a separate image pipeline. The skill ships TypeScript client logic that walks Gemini web response batches, extracts generated image URLs from nested candidate structures, and validates that behavior with unit tests. It depends on baoyu-chrome-cdp for browser control, which implies local Chrome, network access, and comfort running Node scripts from the repo. Treat it as a pragmatic integration layer: powerful for rapid creative iteration, but inherently tied to unofficial web flows that can change without notice. Review Security Audits on the Prism page before installing in production repos.21kinstalls13Baoyu Format MarkdownBaoyu Format Markdown is an agent skill aimed at solo builders who ship technical blogs, product updates, and growth content in markdown. It bundles a title reference with eight proven hook patterns—from subversive takes and concrete numbers to empathy-driven pain-point openers—plus a straightforward headline style when you want clarity over drama. Invoke it while outlining a post, rewriting a draft title, or standardizing how you structure sections before publishing to Dev.to, a docs site, or a landing-page changelog. It does not replace your voice; it gives repeatable formulas so you spend less time staring at a blank H1 and more time distributing work you already built. Best alongside Launch distribution when you need social-ready headlines, or Validate landing copy when a single sentence must carry the value prop.20.5kinstalls14Release Skillsrelease-skills is a universal agent workflow for solo and indie builders who are ready to ship a new version instead of improvising git tags and changelogs. It listens for release-oriented phrases—including 发布, bump version, GitHub Release, and backfill—and walks through detection of the right version file, changelog updates, and push-oriented steps for Node.js, Python, Rust, Claude plugins, and generic layouts. The skill standardizes how you ask the human for choices: prefer built-in clarification tools, fall back to numbered questions, and batch questions when the runtime allows. That matters when one person owns product, code, and ops: fewer missed steps mean fewer broken tags or half-published GitHub Releases. Use it at the end of a feature cycle when manifests and notes should match what actually ships.19.5kinstalls15Baoyu Translatebaoyu-translate is an agent skill for turning English or mixed-language source material into a chosen target language while preserving terminology and tone. Solo builders use it when they need to localize documentation, blog posts, or in-app strings without ad-hoc paste-and-translate in chat. Configuration lives in EXTEND.md: target language, quick versus normal versus refined passes, audience and style (storytelling, formal, technical, literal, and more), and glossary entries merged from inline lists or external markdown or YAML tables. Long inputs split automatically around the chunk threshold so agents do not truncate mid-document. Pair it with your repo’s glossary for product names and AI terms you want kept in English. It fits Claude Code, Cursor, and Codex workflows where the deliverable is finished translated markdown or copy ready to commit.17.1kinstalls16Baoyu Post To Weibobaoyu-post-to-weibo is an agent skill that posts prepared content to Weibo by driving a real browser session and rendering markdown through the baoyu-md stack. Indie builders targeting Chinese distribution use it to turn release notes, changelog snippets, or launch threads into Weibo updates without leaving Claude Code, Cursor, or Codex. The package is a Node workspace that depends on baoyu-chrome-cdp for automation and baoyu-md for structured text, which means you need local Chrome, network access, and logged-in Weibo credentials in the automated profile. It is best treated as a workflow accelerator for repeatable launches—not a replacement for Weibo’s official APIs or compliance review of promotional content. Check Prism Security Audits before installing, and keep human review on copy and attached media before any automated post goes live.14.8kinstalls17Baoyu Youtube TranscriptBaoyu YouTube Transcript is an agent skill that downloads subtitles, metadata, and cover images from YouTube videos using URL or video ID. It targets solo builders and creators who need quotable text from talks, tutorials, or competitor videos without manual copying or a YouTube Data API key. The workflow uses YouTube’s InnerTube path and falls back to yt-dlp when direct access is blocked, supporting multiple languages, translation, chapters, and speaker-aware formatting. Cached raw payloads let you re-export to markdown or other formats quickly. Typical uses span Idea research (mining expert videos), Validate (checking how others explain a problem), and Grow content (repurposing transcripts into posts or docs). Installation expects bun or npx in PATH. It is an integration-style skill with shell scripts, not a hosted MCP server.12.2kinstalls18Baoyu ImagineBaoyu Imagine is an integration-style agent skill for solo builders who need production images without leaving Claude Code, Cursor, or Codex. It wires official APIs across OpenAI, Azure OpenAI, Google, OpenRouter, DashScope, Z.AI GLM-Image, MiniMax, Jimeng, Seedream, and Replicate, supporting text-to-image, reference-guided generation, aspect ratios, and throughput-friendly batch modes when you already have prompt files. Invoke it whenever someone asks to generate, create, or draw images for landing pages, social posts, app store shots, or documentation. It spans multiple journey phases—Build for product and doc visuals, Launch and Grow for distribution creative—while staying a task integration rather than a full creative methodology. Runtime expects bun or npx and handles provider choice through structured user prompts, which keeps API keys and model selection explicit for one-person shops juggling several vendors.9.4kinstalls19Baoyu Diagrambaoyu-diagram is an agent skill that teaches a repeatable layout system for software architecture diagrams. Solo and indie builders use it when they need a readable system map for a README, validation prototype, or production runbook—not a one-off sketch in chat. The skill commits you to one primary flow direction: left-to-right for request and data pipelines (users on the left, stores on the right) or top-to-bottom for classic layered stacks (clients up top, infrastructure below). From there, a five-step procedure groups components by role, assigns one layer per column or row, enforces at least 40px between siblings inside a layer, wraps shared infrastructure in labeled regions, and routes arrows and bus-style connectors across gaps. Included ASCII templates show typical column x-positions and row y-positions so agents place boxes consistently. It does not replace threat modeling or capacity planning; it makes the visual grammar of your architecture documentation predictable so teammates and future you can scan flows in seconds.7.1kinstalls20Baoyu Image CardsBaoyu Image Cards is an agent skill that decomposes dense copy into a cohesive infographic card series tuned for Chinese and international social platforms. Solo builders and indie marketers use it when they already have a message—product launch notes, tutorial steps, or listicle bullets—and need carousel-ready visuals without hiring a designer. The workflow walks through style, layout, and palette selection (twelve styles, eight layouts, three palettes), then generates one to ten cartoon-style frames optimized for saves and shares. It fits the Launch and Grow phases whenever distribution depends on visual snippets rather than a single hero image. The skill expects an agent runtime that can clarify preferences and call an image-generation backend; it does not replace brand strategy or legal review of claims in each card.6.3kinstalls21Baoyu Wechat Summarybaoyu-wechat-summary is an agent skill for turning WeChat history into structured digests you can archive and reread. It centers on a layered EXTEND.md preference file so project-local, XDG, or home config resolves in priority order without hard-coding secrets in the repo. You set your wxid and display name so your own messages render cleanly, choose normal versus roast tone or both, and define default windows such as 24h or 7d when prompts lack explicit dates. Optional data_root keeps digests in a shared folder for solo builders who track communities, clients, or family groups across projects. The skill fits operators who already use wx-style CLI contact queries and want repeatable summaries instead of scrolling raw chat logs.2.9kinstalls