jimliu/baoyu-skills▌

Official API-based image generation. Supports OpenAI, Azure OpenAI, Google, OpenRouter, DashScope (阿里通义万象), MiniMax, Jimeng (即梦), Seedream (豆包) and Replicate providers.

$22

Reverse-engineered Gemini Web API for text and image generation with multi-turn conversation support. \n \n Supports text generation, image generation from prompts, and vision input via reference images for analysis or variation \n Includes four model options: gemini-3-pro (default), gemini-3-flash, gemini-3-flash-thinking, and gemini-3.1-pro-preview \n Multi-turn conversation via session IDs with automatic cookie caching and browser-based Google authentication \n Requires user consent before fi

Professional infographics combining 21 layout types with 20 visual styles for publication-ready output. \n \n Choose from 21 layouts (bento-grid, hierarchical-layers, timeline, funnel, venn-diagram, etc.) and 20 styles (craft-handmade, cyberpunk-neon, technical-schematic, pixel-art, etc.) to match your content structure and visual tone \n Analyzes input content, recommends layout×style combinations, and generates structured markdown before image creation \n Supports custom aspect ratios (landsca

Multi-provider AI image generation with text-to-image, reference images, batch processing, and quality presets. \n \n Supports seven providers: OpenAI, Google, OpenRouter, DashScope, Jimeng, Seedream, and Replicate with automatic provider selection or explicit override \n Handles single images via CLI flags or batch parallel generation from JSON files with configurable worker counts and per-provider throttling \n Supports reference images (Google, OpenAI, OpenRouter, Replicate, Seedream), custom

Three-mode translation with terminology consistency and publication-quality refinement workflows. \n \n Supports quick (direct), normal (analyze then translate), and refined (analyze, translate, review, polish) modes; auto-detects mode from user intent keywords \n Handles long content via intelligent markdown-aware chunking with parallel subagent translation and shared glossary enforcement across chunks \n Customizable via EXTEND.md for target language, mode, audience, style presets (storytellin

Publish articles and image-text posts to WeChat Official Accounts via API or browser automation. \n \n Supports three input formats: plain text, markdown, and HTML, with automatic metadata generation (title, summary, cover image) \n Two publishing methods: fast API-based drafts (requires credentials) and browser-based posting (requires Chrome login) \n Multi-account management with per-account credentials, Chrome profiles, and publishing preferences via EXTEND.md configuration \n Markdown articl

Post text, images, videos, and long-form articles to X via real Chrome browser automation. \n \n Supports four post types: regular posts with images, video posts, quote tweets, and long-form Markdown articles (X Premium required) \n Uses real Chrome with Chrome DevTools Protocol to bypass anti-bot detection; session persists after first manual login \n Scripts fill content into the browser compose window; user reviews and publishes manually \n Requires bun runtime and Google Chrome or Chromium;

Fetch any URL and convert to clean markdown using Chrome CDP with intelligent fallback conversion. \n \n Supports two capture modes: auto-capture on page load or wait-for-user-signal for login-required and lazy-loading pages \n Saves rendered HTML snapshot alongside markdown output with YAML front matter including metadata, URL, title, and capture timestamp \n Upgraded Defuddle-based conversion pipeline with automatic fallback to legacy HTML-to-Markdown extractor; falls back to hosted defuddle.m

Generate customizable article cover images across 5 independent dimensions and 3 aspect ratios. \n \n Combines 6 image types (hero, conceptual, typography, metaphor, scene, minimal) with 10 color palettes and 7 rendering styles for fine-grained visual control \n Supports cinematic (2.35:1), widescreen (16:9), and square (1:1) aspects, plus additional ratios (4:3, 3:2, 3:4) \n Auto-analyzes article content to recommend dimensions, or accepts explicit flags for type, palette, rendering, text level

Transform content into professional slide deck images with customizable styles and audience targeting. \n \n Generates slide outlines with style instructions, then creates individual slide images in 10+ preset styles (blueprint, corporate, sketch-notes, minimal, etc.) or custom dimension combinations \n Supports audience targeting (beginners, executives, experts), language selection, and configurable slide counts (5-30 slides based on content length) \n Includes two-round confirmation workflow w

Post text, images, videos, and Markdown articles to Weibo via Chrome automation. \n \n Two posting modes: regular posts (text with up to 18 images/videos) and long-form headline articles with Markdown input \n Supports custom Chrome profiles and persists login sessions across runs \n Article publishing includes title/summary validation (32 and 44 character limits), cover image support, and automatic image placeholder replacement \n Scripts fill content into the browser; user reviews and manually

Converts Markdown to styled HTML with WeChat-optimized themes, code highlighting, and optional bottom citations. \n \n Supports four themes (default, grace, simple, modern) with customizable colors, fonts, and sizes via CLI options \n Handles advanced Markdown features including syntax-highlighted code blocks, tables, Mermaid/PlantUML diagrams, alerts, footnotes, and ruby text \n Optional citation mode converts external links to numbered references with a bottom \"引用链接\" section, preserving WeCh

Educational comics with flexible art styles, tones, and panel layouts for knowledge storytelling. \n \n Supports five art styles (ligne-claire, manga, realistic, ink-brush, chalk) and seven tones (neutral, warm, dramatic, romantic, energetic, vintage, action) with compatibility guidance to avoid mismatched combinations \n Includes three preset shortcuts (ohmsha for tutorials, wuxia for martial arts, shoujo for romance) that apply special visual rules beyond style selection \n Auto-detects optima

Automated multi-language release workflow with auto-detected version files and semantic versioning. \n \n Supports Node.js, Python, Rust, Claude Plugin, and generic projects with automatic configuration detection \n Generates changelogs in multiple languages (English, Chinese, Japanese, Korean, German, French, Spanish) with proper section translations and third-party contributor attribution \n Groups commits by skill/module, creates separate conventional commits per group, and updates README fil

Downloads transcripts (subtitles/captions) from YouTube videos. Works with both manually created and auto-generated transcripts. No API key or browser required — uses YouTube's InnerTube API directly and automatically falls back to yt-dlp when YouTube blocks the direct API path.

Transforms plain text or markdown into well-structured, reader-friendly markdown with improved formatting. \n \n Analyzes content structure, identifies key insights, and applies strategic formatting (headings, bold, lists, tables, blockquotes) without changing original words \n Generates or optimizes frontmatter with title, slug, summary, and description fields; supports auto-selection via EXTEND.md configuration \n Runs typography scripts for CJK/English spacing, emphasis fixes, and optional qu

Generates cartoon-style infographic series for Xiaohongshu with 11 visual styles and 8 layout options. \n \n Breaks content into 1-10 images optimized for XHS engagement, with auto-detection of content type (reviews, tutorials, lifestyle, knowledge cards) \n Offers 11 distinct visual styles (cute, notion, chalkboard, screen-print, retro, bold, minimal, and others) freely combinable with 8 layout structures (sparse, dense, list, comparison, flow, mindmap, quadrant) \n Provides preset shortcuts (k

Convert X tweets and articles to markdown with YAML front matter using reverse-engineered API. \n \n Extracts tweets, threads, and X articles to markdown with metadata (author, URL, cover image, thread count) \n Requires explicit user consent before first use; displays disclaimer about reverse-engineered API and potential account risks \n Supports media download to local imgs/ and videos/ directories with automatic link rewriting, configurable per-conversion or via preferences \n Authenticates v

Compresses images to WebP or PNG with automatic tool selection based on system availability. \n \n Supports WebP (default), PNG, and JPEG output formats with configurable quality (0–100, default 80) \n Automatically selects best available compression tool: sips, cwebp, ImageMagick, or Sharp \n Processes single files or directories recursively with options to keep originals or replace in-place \n Customizable via EXTEND.md configuration files at project, user home, or XDG config locations \n