chore: release v1.4.0

chore: release v1.3.0
chore: release v1.2.0
2026-07-10 13:11:30 +00:00 · 2026-01-18 23:46:07 -06:00 · 2026-01-18 22:16:24 -06:00 · 2026-01-18 20:42:40 -06:00 · 2026-01-18 20:28:36 -06:00 · 2026-01-18 20:11:49 -06:00
170 changed files with 11817 additions and 3178 deletions
@@ -6,16 +6,15 @@
  },
  "metadata": {
    "description": "Skills shared by Baoyu for improving daily work efficiency",
-    "version": "0.5.2"
+    "version": "1.4.0"
  },
  "plugins": [
    {
      "name": "content-skills",
-      "description": "Skills shared by Baoyu for improving daily work efficiency",
+      "description": "Content generation and publishing skills",
      "source": "./",
      "strict": false,
      "skills": [
-        "./skills/baoyu-gemini-web",
        "./skills/baoyu-xhs-images",
        "./skills/baoyu-post-to-x",
        "./skills/baoyu-post-to-wechat",
@@ -24,6 +23,25 @@
        "./skills/baoyu-slide-deck",
        "./skills/baoyu-comic"
      ]
+    },
+    {
+      "name": "ai-generation-skills",
+      "description": "AI-powered generation backends",
+      "source": "./",
+      "strict": false,
+      "skills": [
+        "./skills/baoyu-danger-gemini-web"
+      ]
+    },
+    {
+      "name": "utility-skills",
+      "description": "Utility tools for content processing",
+      "source": "./",
+      "strict": false,
+      "skills": [
+        "./skills/baoyu-danger-x-to-markdown",
+        "./skills/baoyu-compress-image"
+      ]
    }
  ]
 }
@@ -0,0 +1,203 @@
+---
+name: release-skills
+description: Release workflow for baoyu-skills plugin. This skill should be used when the user wants to create a new release version. It analyzes changes since the last version tag, updates changelogs (EN/CN), bumps the version in marketplace.json, commits changes, and creates a version tag. Supports dry-run mode and breaking change detection.
+---
+
+# Release Skills
+
+Automate the release process for baoyu-skills plugin: analyze changes, update changelogs, bump version, commit, and tag.
+
+## When to Use
+
+Trigger this skill when user requests:
+- "release", "发布", "create release", "new version"
+- "bump version", "update version"
+- "prepare release"
+
+## Workflow
+
+### Step 1: Analyze Changes Since Last Tag
+
+```bash
+# Get the latest version tag
+LAST_TAG=$(git tag --sort=-v:refname | head -1)
+
+# Show changes since last tag
+git log ${LAST_TAG}..HEAD --oneline
+git diff ${LAST_TAG}..HEAD --stat
+```
+
+Categorize changes by type based on commit messages and file changes:
+
+| Type | Prefix | Description |
+|------|--------|-------------|
+| feat | `feat:` | New features, new skills |
+| fix | `fix:` | Bug fixes |
+| docs | `docs:` | Documentation only |
+| refactor | `refactor:` | Code refactoring |
+| style | `style:` | Formatting, styling |
+| chore | `chore:` | Build, tooling, maintenance |
+
+**Breaking Change Detection**: If changes include:
+- Removed skills or scripts
+- Changed API/interfaces
+- Renamed public functions/options
+
+Warn user: "Breaking changes detected. Consider major version bump (--major flag)."
+
+### Step 2: Determine Version Bump
+
+Current version location: `.claude-plugin/marketplace.json` → `metadata.version`
+
+Version rules:
+- **Patch** (0.6.1 → 0.6.2): Bug fixes, docs updates, minor improvements
+- **Minor** (0.6.x → 0.7.0): New features, new skills, significant enhancements
+- **Major** (0.x → 1.0): Breaking changes, only when user explicitly requests with `--major`
+
+Default behavior:
+- If changes include `feat:` or new skills → Minor bump
+- Otherwise → Patch bump
+
+### Step 3: Check and Update README
+
+Before updating changelogs, check if README files need updates based on changes:
+
+**When to update README**:
+- New skills added → Add to skill list
+- Skills removed → Remove from skill list
+- Skill renamed → Update references
+- New features affecting usage → Update usage section
+- Breaking changes → Update migration notes
+
+**Files to sync**:
+- `README.md` (English)
+- `README.zh.md` (Chinese)
+
+If changes include new skills or significant feature changes, update both README files to reflect the new capabilities. Keep both files in sync with the same structure and information.
+
+### Step 4: Update Changelogs
+
+Files to update:
+- `CHANGELOG.md` (English)
+- `CHANGELOG.zh.md` (Chinese)
+
+Format (insert after header, before previous version):
+
+```markdown
+## {NEW_VERSION} - {YYYY-MM-DD}
+
+### Features
+- `skill-name`: description of new feature
+
+### Fixes
+- `skill-name`: description of fix
+
+### Documentation
+- description of docs changes
+
+### Other
+- description of other changes
+```
+
+Only include sections that have changes. Omit empty sections.
+
+For Chinese changelog, translate the content maintaining the same structure.
+
+### Step 5: Update marketplace.json
+
+Update `.claude-plugin/marketplace.json`:
+```json
+{
+  "metadata": {
+    "version": "{NEW_VERSION}"
+  }
+}
+```
+
+### Step 6: Commit Changes
+
+```bash
+git add README.md README.zh.md CHANGELOG.md CHANGELOG.zh.md .claude-plugin/marketplace.json
+git commit -m "chore: release v{NEW_VERSION}"
+```
+
+**Note**: Do NOT add Co-Authored-By line. This is a release commit, not a code contribution.
+
+### Step 7: Create Version Tag
+
+```bash
+git tag v{NEW_VERSION}
+```
+
+**Important**: Do NOT push to remote. User will push manually when ready.
+
+## Options
+
+| Flag | Description |
+|------|-------------|
+| `--dry-run` | Preview changes without executing. Show what would be updated. |
+| `--major` | Force major version bump (0.x → 1.0 or 1.x → 2.0) |
+| `--minor` | Force minor version bump |
+| `--patch` | Force patch version bump |
+| `--pre <tag>` | (Reserved) Create pre-release version, e.g., `--pre beta` → `0.7.0-beta.1` |
+
+## Dry-Run Mode
+
+When `--dry-run` is specified:
+1. Show all changes since last tag
+2. Show proposed version bump (current → new)
+3. Show draft changelog entries (EN and CN)
+4. Show files that would be modified
+5. Do NOT make any actual changes
+
+Output format:
+```
+=== DRY RUN MODE ===
+
+Last tag: v0.6.1
+Proposed version: v0.7.0
+
+Changes detected:
+- feat: new skill baoyu-foo added
+- fix: baoyu-bar timeout issue
+- docs: updated README
+
+Changelog preview (EN):
+## 0.7.0 - 2026-01-17
+### Features
+- `baoyu-foo`: new skill for ...
+### Fixes
+- `baoyu-bar`: fixed timeout issue
+
+README updates needed: Yes/No
+(If yes, show proposed changes)
+
+Files to modify:
+- README.md (if updates needed)
+- README.zh.md (if updates needed)
+- CHANGELOG.md
+- CHANGELOG.zh.md
+- .claude-plugin/marketplace.json
+
+No changes made. Run without --dry-run to execute.
+```
+
+## Example Usage
+
+```
+/release-skills              # Auto-detect version bump
+/release-skills --dry-run    # Preview only
+/release-skills --minor      # Force minor bump
+/release-skills --major      # Force major bump (with confirmation)
+```
+
+## Post-Release Reminder
+
+After successful release, remind user:
+```
+Release v{NEW_VERSION} created locally.
+
+To publish:
+  git push origin main
+  git push origin v{NEW_VERSION}
+```
@@ -140,3 +140,7 @@ vite.config.ts.timestamp-*
 tests-data/

 .DS_Store
+
+# Skill extensions (user customization)
+.baoyu-skills/
+x-to-markdown/
@@ -0,0 +1,197 @@
+# Changelog
+
+English | [中文](./CHANGELOG.zh.md)
+
+## 1.4.0 - 2026-01-18
+
+### Features
+- `baoyu-cover-image`: expands style library from 8 to 19 styles with 12 new additions—`blueprint`, `bold-editorial`, `chalkboard`, `dark-atmospheric`, `editorial-infographic`, `fantasy-animation`, `intuition-machine`, `notion`, `pixel-art`, `sketch-notes`, `vector-illustration`, `vintage`, `watercolor`.
+- `baoyu-slide-deck`: adds `chalkboard` style—black chalkboard background with colorful chalk drawings for education and tutorials.
+
+### Breaking Changes
+- `baoyu-cover-image`: removes `tech` style (use `blueprint` or `editorial-infographic` for technical content).
+
+### Documentation
+- `README.md`, `README.zh.md`: updates style preview screenshots for cover-image and slide-deck.
+
+## 1.3.0 - 2026-01-18
+
+### Features
+- `baoyu-comic`: adds `wuxia` style—Hong Kong martial arts comic style with ink brush strokes, dynamic combat poses, and qi energy effects. Best for wuxia/xianxia and Chinese historical fiction.
+- `baoyu-comic`: adds style and layout preview screenshots for all 8 styles and 6 layouts in README.
+
+### Refactor
+- `baoyu-comic`: removes `tech` style (replaced by `ohmsha` for technical content).
+
+## 1.2.0 - 2026-01-18
+
+### Features
+- Session-independent output directories: each generation session creates a new directory (`<skill-suffix>/<topic-slug>/`), even for the same source file. Conflicts resolved by appending timestamp.
+- Multi-source file support: source files now saved as `source-{slug}.{ext}`, supporting multiple inputs (text, images, files from conversation).
+
+### Documentation
+- `CLAUDE.md`: updates Output Path Convention with new session-independent directory structure and multi-source file naming.
+- Multiple skills: updates file management sections to reflect new directory and source file conventions.
+  - `baoyu-slide-deck`, `baoyu-article-illustrator`, `baoyu-cover-image`, `baoyu-xhs-images`, `baoyu-comic`
+
+## 1.1.0 - 2026-01-18
+
+### Features
+- `baoyu-compress-image`: new utility skill for cross-platform image compression. Converts to WebP by default with PNG-to-PNG support. Uses system tools (sips, cwebp, ImageMagick) with Sharp fallback.
+
+### Refactor
+- Marketplace structure: reorganizes plugins into three categories—`content-skills`, `ai-generation-skills`, and `utility-skills`—for better organization.
+
+### Documentation
+- `CLAUDE.md`, `README.md`, `README.zh.md`: updates skill architecture documentation to reflect the new three-category structure.
+
+## 1.0.1 - 2026-01-18
+
+### Refactor
+- Code structure improvements for better readability and maintainability.
+- `baoyu-slide-deck`: unified style reference file formats.
+
+### Other
+- Screenshots: converted from PNG to WebP format for smaller file sizes; added screenshots for new styles.
+
+## 1.0.0 - 2026-01-18
+
+### Features
+- `baoyu-danger-x-to-markdown`: new skill to convert X/Twitter posts and threads to Markdown format.
+
+### Breaking Changes
+- `baoyu-gemini-web` renamed to `baoyu-danger-gemini-web` to indicate potential risks of using reverse-engineered APIs.
+
+## 0.11.0 - 2026-01-18
+
+### Features
+- `baoyu-danger-gemini-web`: adds disclaimer consent check flow—requires user acceptance before first use, with persistent consent storage per platform.
+
+## 0.10.0 - 2026-01-18
+
+### Features
+- `baoyu-slide-deck`: expands style library from 10 to 15 styles with 8 new additions—`dark-atmospheric`, `editorial-infographic`, `fantasy-animation`, `intuition-machine`, `pixel-art`, `scientific`, `vintage`, `watercolor`.
+
+### Breaking Changes
+- `baoyu-slide-deck`: removes 3 styles (`playful`, `storytelling`, `warm`); changes default style from `notion` to `blueprint`.
+
+## 0.9.0 - 2026-01-17
+
+### Features
+- Extension support: all skills now support customization via `EXTEND.md` files. Check `.baoyu-skills/<skill-name>/EXTEND.md` (project) or `~/.baoyu-skills/<skill-name>/EXTEND.md` (user) for custom styles and configurations.
+
+### Other
+- `.gitignore`: adds `.baoyu-skills/` directory for user extension files.
+
+## 0.8.2 - 2026-01-17
+
+### Refactor
+- `baoyu-danger-gemini-web`: reorganizes script architecture—moves modular files into `gemini-webapi/` subdirectory and updates SKILL.md with `${SKILL_DIR}` path references.
+
+## 0.8.1 - 2026-01-17
+
+### Refactor
+- `baoyu-danger-gemini-web`: refactors script architecture—consolidates 10 separate files into a structured `gemini-webapi/` module (TypeScript port of gemini_webapi Python library).
+
+## 0.8.0 - 2026-01-17
+
+### Features
+- `baoyu-xhs-images`: adds content analysis framework (`analysis-framework.md`, `outline-template.md`) for structured content breakdown and outline generation.
+
+### Documentation
+- `CLAUDE.md`: adds Output Path Convention (directory structure, backup rules) and Image Naming Convention (format, slug rules) to standardize image generation outputs.
+- Multiple skills: updates file management conventions to use unified directory structure (`[source-name-no-ext]/<skill-suffix>/`).
+  - `baoyu-article-illustrator`, `baoyu-comic`, `baoyu-cover-image`, `baoyu-slide-deck`, `baoyu-xhs-images`
+
+## 0.7.0 - 2026-01-17
+
+### Features
+- `baoyu-comic`: adds `--aspect` (3:4, 4:3, 16:9) and `--lang` options; introduces multi-variant storyboard workflow (chronological, thematic, character-centric) with user selection.
+
+### Enhancements
+- `baoyu-comic`: adds `analysis-framework.md` and `storyboard-template.md` for structured content analysis and variant generation.
+- `baoyu-slide-deck`: adds `analysis-framework.md`, `content-rules.md`, `modification-guide.md`, and `outline-template.md` references for improved outline quality.
+- `baoyu-article-illustrator`, `baoyu-cover-image`, `baoyu-xhs-images`: enhanced SKILL.md documentation with clearer workflows.
+
+### Documentation
+- Multiple skills: restructured SKILL.md files—moved detailed content to `references/` directory for maintainability.
+- `baoyu-slide-deck`: simplified SKILL.md, consolidated style descriptions.
+
+## 0.6.1 - 2026-01-17
+
+- `baoyu-slide-deck`: adds `scripts/merge-to-pdf.ts` to export generated slides into a single PDF; docs updated with pptx/pdf outputs.
+- `baoyu-comic`: adds `scripts/merge-to-pdf.ts` to merge cover/pages into a PDF; docs clarify character reference handling (image vs text).
+- Docs conventions: adds a “Script Directory” template to `CLAUDE.md`; aligns `baoyu-danger-gemini-web` / `baoyu-slide-deck` / `baoyu-comic` docs to use `${SKILL_DIR}` in commands so agents can run scripts from any install location.
+
+## 0.6.0 - 2026-01-17
+
+- `baoyu-slide-deck`: adds `scripts/merge-to-pptx.ts` to merge slide images into a PPTX and attach `prompts/` content as speaker notes.
+- `baoyu-slide-deck`: reshapes/expands the style library (adds `blueprint` / `bold-editorial` / `sketch-notes` / `vector-illustration`, and adjusts/replaces some older styles).
+- `baoyu-comic`: adds a `realistic` style reference.
+- Docs: refreshes `README.md` / `README.zh.md`.
+
+## 0.5.3 - 2026-01-17
+
+- `baoyu-post-to-x` (X Articles): makes image placeholder replacement more reliable (selection retry + verification; deletes via Backspace and verifies deletion before pasting), reducing mis-insertions/failures.
+
+## 0.5.2 - 2026-01-16
+
+- `baoyu-danger-gemini-web`: adds `--sessionId` (local persisted sessions, plus `--list-sessions`) for multi-turn conversations and consistent multi-image generation.
+- `baoyu-danger-gemini-web`: adds `--reference/--ref` for reference images (vision input), plus stronger timeout handling and cookie refresh recovery.
+- Docs: `baoyu-xhs-images` / `baoyu-slide-deck` / `baoyu-comic` document session usage (reuse one `sessionId` per set) to improve visual consistency.
+
+## 0.5.1 - 2026-01-16
+
+- `baoyu-comic`: adds creation templates/references (character template, Ohmsha guide, outline template) to speed up “characters → storyboard → generation”.
+
+## 0.5.0 - 2026-01-16
+
+- Adds `baoyu-comic`: a knowledge-comic generator with `style × layout` and a full set of style/layout references for more stable output.
+- `baoyu-xhs-images`: moves style/layout details into `references/styles/*` and `references/layouts/*`, and migrates the base prompt into `references/base-prompt.md` for easier maintenance/reuse.
+- `baoyu-slide-deck` / `baoyu-cover-image`: similarly split base prompt and style references into `references/`, reducing SKILL.md complexity and making style expansion easier.
+- Docs: updates `README.md` / `README.zh.md` skill list and examples.
+
+## 0.4.2 - 2026-01-15
+
+- `baoyu-danger-gemini-web`: updates description to clarify it as the image-generation backend for other skills (e.g. `cover-image`, `xhs-images`, `article-illustrator`).
+
+## 0.4.1 - 2026-01-15
+
+- `baoyu-post-to-x` / `baoyu-post-to-wechat`: adds `scripts/paste-from-clipboard.ts` to send a “real paste” keystroke (Cmd/Ctrl+V), avoiding sites ignoring CDP synthetic events.
+- `baoyu-post-to-x`: adds docs for X Articles/regular posts, and switches image upload to prefer real paste (with a CDP fallback).
+- `baoyu-post-to-wechat`: docs add script-location guidance and `${SKILL_DIR}` path usage for reliable agent execution.
+- Docs: adds `screenshots/update-plugins.png` for the marketplace update flow.
+
+## 0.4.0 - 2026-01-15
+
+- Adds `baoyu-` prefix to skill directories and updates marketplace paths/docs accordingly to reduce naming collisions.
+
+## 0.3.1 - 2026-01-15
+
+- `xhs-images`: upgrades docs to a Style × Layout system (adds `--layout`, auto layout selection, and a `notion` style), with more complete usage examples.
+- `article-illustrator` / `cover-image`: docs no longer hard-code `gemini-web`; instead they instruct the agent to pick an available image-generation skill.
+- `slide-deck`: docs add the `notion` style and update auto-style mapping.
+- Tooling/docs: adds `.DS_Store` to `.gitignore`; refreshes `README.md` / `README.zh.md`.
+
+## 0.3.0 - 2026-01-14
+
+- Adds `post-to-wechat`: Chrome CDP automation for WeChat Official Account posting (image-text + full article), including Markdown → WeChat HTML conversion and multiple themes.
+- Adds `CLAUDE.md`: repository structure, running conventions, and “add new skill” guidelines.
+- Docs: updates `README.md` / `README.zh.md` install/update/usage instructions.
+
+## 0.2.0 - 2026-01-13
+
+- Adds new skills: `post-to-x` (real Chrome/CDP automation for posts and X Articles), `article-illustrator`, `cover-image`, and `slide-deck`.
+- `xhs-images`: adds multi-style support (`--style`) with auto style selection and updates the base prompt (e.g. language follows input, hand-drawn infographic constraints).
+- Docs: adds `README.zh.md` and improves `README.md` and `.gitignore`.
+
+## 0.1.1 - 2026-01-13
+
+- Marketplace refactor: introduces `metadata` (including `version`), renames the plugin entry to `content-skills` and explicitly lists installable skills; removes legacy `.claude-plugin/plugin.json`.
+- Adds `xhs-images`: Xiaohongshu infographic series generator (outline + per-image prompts).
+- `gemini-web`: adds `--promptfiles` to build prompts from multiple files (system/content separation).
+- Docs: adds `README.md`.
+
+## 0.1.0 - 2026-01-13
+
+- Initial release: `.claude-plugin/marketplace.json` plus `gemini-web` (text/image generation, browser login + cookie cache).
@@ -0,0 +1,196 @@
+# Changelog
+
+[English](./CHANGELOG.md) | 中文
+
+## 1.4.0 - 2026-01-18
+
+### 新功能
+- `baoyu-cover-image`：风格库从 8 个扩展至 19 个，新增 12 种风格——`blueprint`（蓝图）、`bold-editorial`（大胆编辑）、`chalkboard`（黑板）、`dark-atmospheric`（暗黑氛围）、`editorial-infographic`（杂志信息图）、`fantasy-animation`（奇幻动画）、`intuition-machine`（技术简报）、`notion`（Notion 风格）、`pixel-art`（像素艺术）、`sketch-notes`（手绘笔记）、`vector-illustration`（矢量插画）、`vintage`（复古文献）、`watercolor`（水彩）。
+- `baoyu-slide-deck`：新增 `chalkboard`（黑板）风格——黑色黑板背景配彩色粉笔绘画，适合教育和教程内容。
+
+### 破坏性变更
+- `baoyu-cover-image`：移除 `tech` 风格（技术内容改用 `blueprint` 或 `editorial-infographic` 风格）。
+
+### 文档
+- `README.md`、`README.zh.md`：更新 cover-image 和 slide-deck 风格预览截图。
+
+## 1.3.0 - 2026-01-18
+
+### 新功能
+- `baoyu-comic`：新增 `wuxia` 武侠风格——港漫武侠风格，水墨笔触、动态打斗、气功特效。适用于武侠、仙侠、中国历史小说。
+- `baoyu-comic`：README 新增风格和布局预览截图（8 种风格 + 6 种布局）。
+
+### 重构
+- `baoyu-comic`：移除 `tech` 风格（技术内容改用 `ohmsha` 风格）。
+
+## 1.2.0 - 2026-01-18
+
+### 新功能
+- Session 独立输出目录：每次生成创建独立目录（`<skill-suffix>/<topic-slug>/`），即使是同一源文件也会新建目录。目录冲突时追加时间戳。
+- 多源文件支持：源文件现以 `source-{slug}.{ext}` 命名，支持多个输入（文本、图片、会话中的文件）。
+
+### 文档
+- `CLAUDE.md`：更新 Output Path Convention，采用新的 session 独立目录结构和多源文件命名规范。
+- 多个技能：更新文件管理部分，反映新的目录和源文件规范。
+  - `baoyu-slide-deck`、`baoyu-article-illustrator`、`baoyu-cover-image`、`baoyu-xhs-images`、`baoyu-comic`
+
+## 1.1.0 - 2026-01-18
+
+### 新功能
+- `baoyu-compress-image`：新增跨平台图片压缩技能。默认转换为 WebP 格式，支持 PNG 转 PNG。自动选择系统工具（sips、cwebp、ImageMagick），Sharp 作为兜底方案。
+
+### 重构
+- Marketplace 结构重组：将插件分为三大类——`content-skills`（内容技能）、`ai-generation-skills`（AI 生成技能）和 `utility-skills`（工具技能），便于管理和发现。
+
+### 文档
+- `CLAUDE.md`、`README.md`、`README.zh.md`：更新技能架构文档，反映新的三类分组结构。
+
+## 1.0.1 - 2026-01-18
+
+### 重构
+- 代码结构优化，提升可读性和可维护性。
+- `baoyu-slide-deck`：统一风格参考文件格式。
+
+### 其他
+- 截图：从 PNG 转换为 WebP 格式，减小文件体积；新增新风格的截图。
+
+## 1.0.0 - 2026-01-18
+
+### 新功能
+- `baoyu-danger-x-to-markdown`：新增技能，将 X/Twitter 帖子和线程转换为 Markdown 格式。
+
+### 破坏性变更
+- `baoyu-gemini-web` 重命名为 `baoyu-danger-gemini-web`，以提示使用逆向工程 API 的潜在风险。
+
+## 0.11.0 - 2026-01-18
+
+### 新功能
+- `baoyu-danger-gemini-web`：新增 Disclaimer 同意检查流程——首次使用前需用户确认接受，同意状态按平台持久化存储。
+
+## 0.10.0 - 2026-01-18
+
+### 新功能
+- `baoyu-slide-deck`：风格库从 10 个扩展至 15 个，新增 8 种风格——`dark-atmospheric`（暗黑氛围）、`editorial-infographic`（杂志信息图）、`fantasy-animation`（奇幻动画）、`intuition-machine`（技术简报）、`pixel-art`（像素艺术）、`scientific`（科学图解）、`vintage`（复古文献）、`watercolor`（水彩手绘）。
+
+### 破坏性变更
+- `baoyu-slide-deck`：移除 3 种风格（`playful`、`storytelling`、`warm`）；默认风格从 `notion` 改为 `blueprint`。
+
+## 0.9.0 - 2026-01-17
+
+### 新功能
+- 扩展支持：所有技能现支持通过 `EXTEND.md` 文件自定义。检查 `.baoyu-skills/<skill-name>/EXTEND.md`（项目级）或 `~/.baoyu-skills/<skill-name>/EXTEND.md`（用户级）配置自定义样式与设置。
+
+### 其他
+- `.gitignore`：添加 `.baoyu-skills/` 目录忽略，存放用户扩展文件。
+
+## 0.8.2 - 2026-01-17
+
+### 重构
+- `baoyu-danger-gemini-web`：重组脚本架构——将模块文件移至 `gemini-webapi/` 子目录，并更新 SKILL.md 使用 `${SKILL_DIR}` 路径引用。
+
+## 0.8.1 - 2026-01-17
+
+### 重构
+- `baoyu-danger-gemini-web`：重构脚本架构——将 10 个分散的脚本文件整合为结构化的 `gemini-webapi/` 模块（gemini_webapi Python 库的 TypeScript 移植版）。
+
+## 0.8.0 - 2026-01-17
+
+### 新功能
+- `baoyu-xhs-images`：新增内容分析框架（`analysis-framework.md`、`outline-template.md`），提供结构化内容拆解与大纲生成方案。
+
+### 文档
+- `CLAUDE.md`：新增 Output Path Convention（目录结构、备份规则）和 Image Naming Convention（文件命名格式、slug 规则），统一图片生成输出规范。
+- 多个技能：更新文件管理规范，采用统一目录结构（`[source-name-no-ext]/<skill-suffix>/`）。
+  - `baoyu-article-illustrator`、`baoyu-comic`、`baoyu-cover-image`、`baoyu-slide-deck`、`baoyu-xhs-images`
+
+## 0.7.0 - 2026-01-17
+
+### 新功能
+- `baoyu-comic`：新增 `--aspect`（3:4、4:3、16:9）和 `--lang` 选项；引入多变体分镜工作流（时间线、主题、人物视角），支持用户选择最佳方案。
+
+### 增强
+- `baoyu-comic`：新增 `analysis-framework.md` 和 `storyboard-template.md`，提供结构化内容分析与变体生成框架。
+- `baoyu-slide-deck`：新增 `analysis-framework.md`、`content-rules.md`、`modification-guide.md`、`outline-template.md` 参考文档，提升大纲质量。
+- `baoyu-article-illustrator`、`baoyu-cover-image`、`baoyu-xhs-images`：SKILL.md 文档增强，工作流程更清晰。
+
+### 文档
+- 多个技能：重构 SKILL.md 结构，将详细内容移至 `references/` 目录，便于维护。
+- `baoyu-slide-deck`：精简 SKILL.md，整合风格描述。
+
+## 0.6.1 - 2026-01-17
+
+- `baoyu-slide-deck`：新增 `scripts/merge-to-pdf.ts`，可将生成的 slide 图片一键合并为 PDF；文档补充导出步骤与产物命名（pptx/pdf）。
+- `baoyu-comic`：新增 `scripts/merge-to-pdf.ts`，将封面/分页图片合并为 PDF；补充角色参考（图片/文本）处理说明。
+- 文档规范：在 `CLAUDE.md` 中补充“Script Directory”模板；`baoyu-danger-gemini-web` / `baoyu-slide-deck` / `baoyu-comic` 文档统一用 `${SKILL_DIR}` 引用脚本路径，方便 agent 在任意安装目录运行。
+
+## 0.6.0 - 2026-01-17
+
+- `baoyu-slide-deck`：新增 `scripts/merge-to-pptx.ts`，将生成的 slide 图片合并为 PPTX，并可把 `prompts/` 写入 speaker notes。
+- `baoyu-slide-deck`：风格库重组与扩充（新增 `blueprint` / `bold-editorial` / `sketch-notes` / `vector-illustration`，并调整/替换部分旧风格定义）。
+- `baoyu-comic`：新增 `realistic` 风格参考文件。
+- 文档：README / README.zh 同步更新技能说明与用法示例。
+
+## 0.5.3 - 2026-01-17
+
+- `baoyu-post-to-x`（X Articles）：插图占位符替换更稳定——选中占位符增加重试与校验，改用 Backspace 删除并确认删除后再粘贴图片，降低插图错位/替换失败概率。
+
+## 0.5.2 - 2026-01-16
+
+- `baoyu-danger-gemini-web`：新增 `--sessionId`（本地持久化会话，支持 `--list-sessions`），用于多轮对话/多图生成保持上下文一致。
+- `baoyu-danger-gemini-web`：新增 `--reference/--ref` 传入参考图片（vision 输入），并增强超时与 cookie 失效自动恢复逻辑。
+- `baoyu-xhs-images` / `baoyu-slide-deck` / `baoyu-comic`：文档补充 session 约定（整套图使用同一 `sessionId`，增强风格一致性）。
+
+## 0.5.1 - 2026-01-16
+
+- `baoyu-comic`：补齐创作模板与参考（角色模板、Ohmsha 教学漫画指南、大纲模板），更适合从“设定 → 分镜 → 生成”快速落地。
+
+## 0.5.0 - 2026-01-16
+
+- 新增 `baoyu-comic`：知识漫画生成器，支持 `style × layout` 组合，并提供风格/布局参考文件用于稳定出图。
+- `baoyu-xhs-images`：将 Style/Layout 的细节从 SKILL.md 拆分到 `references/styles/*` 与 `references/layouts/*`，并将基础提示词迁移到 `references/base-prompt.md`，便于维护和复用。
+- `baoyu-slide-deck` / `baoyu-cover-image`：同样将基础提示词与风格拆分到 `references/`，降低 SKILL.md 复杂度，便于扩展更多风格。
+- 文档：README / README.zh 更新技能清单与用法示例。
+
+## 0.4.2 - 2026-01-15
+
+- `baoyu-danger-gemini-web`：描述信息更新，明确其作为 `cover-image` / `xhs-images` / `article-illustrator` 等技能的图片生成后端。
+
+## 0.4.1 - 2026-01-15
+
+- `baoyu-post-to-x` / `baoyu-post-to-wechat`：新增 `scripts/paste-from-clipboard.ts`，通过系统级 Cmd/Ctrl+V 发送“真实粘贴”按键，规避 CDP 合成事件在站点侧被忽略的问题。
+- `baoyu-post-to-x`：补充 X Articles/普通推文的操作文档（`references/articles.md`、`references/regular-posts.md`），并将发图流程改为优先使用“真实粘贴”（保留 CDP 兜底）。
+- `baoyu-post-to-wechat`：文档补充脚本目录说明与 `${SKILL_DIR}` 路径写法，便于 agent 可靠定位脚本。
+- 文档：新增插件更新流程截图 `screenshots/update-plugins.png`。
+
+## 0.4.0 - 2026-01-15
+
+- 技能命名统一加 `baoyu-` 前缀：目录结构、marketplace 清单与文档示例命令同步更新，减少与其它插件技能的命名冲突。
+
+## 0.3.1 - 2026-01-15
+
+- `xhs-images`：升级为 Style × Layout 二维系统（新增 `--layout`、自动布局选择与 Notion 风格），文档示例更完整。
+- `article-illustrator` / `slide-deck` / `cover-image`：文档改为“选择可用的图片生成技能”而非强绑定 `gemini-web`，并补充 Notion 风格相关说明。
+- 工程化：`.gitignore` 增加 `.DS_Store` 忽略；README / README.zh 同步调整。
+
+## 0.3.0 - 2026-01-14
+
+- 新增 `post-to-wechat`：基于 Chrome CDP 自动化发布公众号图文/文章，包含 Markdown → 微信 HTML 转换与多主题样式支持。
+- 新增 `CLAUDE.md`：补充仓库结构、运行方式与添加新技能的约定，方便协作与二次开发。
+- 文档：README / README.zh 更新安装、更新与使用说明。
+
+## 0.2.0 - 2026-01-13
+
+- 新增技能：`post-to-x`（真实 Chrome/CDP 自动化发布推文与 X Articles）、`article-illustrator`（文章智能插图规划）、`cover-image`（文章封面图生成）、`slide-deck`（幻灯片大纲与图片生成）。
+- `xhs-images`：新增 `--style` 多风格与自动风格选择，并更新基础提示词（例如语言随内容、强调手绘信息图等）。
+- 文档：新增 `README.zh.md`，并完善 README 与 `.gitignore`。
+
+## 0.1.1 - 2026-01-13
+
+- marketplace 结构重构：引入 `metadata`（含 `version`），插件名调整为 `content-skills` 并显式列出可安装 skills；移除旧 `.claude-plugin/plugin.json`。
+- 新增 `xhs-images`：小红书信息图系列生成技能（拆解内容、生成 outline 与提示词）。
+- `gemini-web`：新增 `--promptfiles`，支持从多个文件拼接 prompt（便于 system/content 分离）。
+- 文档：新增 `README.md`。
+
+## 0.1.0 - 2026-01-13
+
+- 初始发布：提供 `.claude-plugin/marketplace.json` 与 `gemini-web`（文本/图片生成、cookie 登录与缓存流程）。
@@ -8,17 +8,34 @@ Claude Code marketplace plugin providing AI-powered content generation skills. S

 ## Architecture

+Skills are organized into three plugin categories in `marketplace.json`:
+
 ```
 skills/
-├── baoyu-gemini-web/          # Core: Gemini API wrapper (text + image gen)
-├── baoyu-xhs-images/          # Xiaohongshu infographic series (1-10 images)
-├── baoyu-cover-image/         # Article cover images (2.35:1 aspect)
-├── baoyu-slide-deck/          # Presentation slides with outlines
-├── baoyu-article-illustrator/ # Smart illustration placement
-├── baoyu-post-to-x/           # X/Twitter posting automation
-└── baoyu-post-to-wechat/      # WeChat Official Account posting
+├── [content-skills]           # Content generation and publishing
+│   ├── baoyu-xhs-images/          # Xiaohongshu infographic series (1-10 images)
+│   ├── baoyu-cover-image/         # Article cover images (2.35:1 aspect)
+│   ├── baoyu-slide-deck/          # Presentation slides with outlines
+│   ├── baoyu-article-illustrator/ # Smart illustration placement
+│   ├── baoyu-comic/               # Knowledge comics (Logicomix/Ohmsha style)
+│   ├── baoyu-post-to-x/           # X/Twitter posting automation
+│   └── baoyu-post-to-wechat/      # WeChat Official Account posting
+│
+├── [ai-generation-skills]     # AI-powered generation backends
+│   └── baoyu-danger-gemini-web/   # Gemini API wrapper (text + image gen)
+│
+└── [utility-skills]           # Utility tools for content processing
+    ├── baoyu-danger-x-to-markdown/ # X/Twitter content to markdown
+    └── baoyu-compress-image/      # Image compression
 ```

+**Plugin Categories**:
+| Category | Description |
+|----------|-------------|
+| `content-skills` | Skills that generate or publish content (images, slides, comics, posts) |
+| `ai-generation-skills` | Backend skills providing AI generation capabilities |
+| `utility-skills` | Helper tools for content processing (conversion, compression) |
+
 Each skill contains:
 - `SKILL.md` - YAML front matter (name, description) + documentation
 - `scripts/` - TypeScript implementations
@@ -35,24 +52,24 @@ npx -y bun skills/<skill>/scripts/main.ts [options]
 Examples:
 ```bash
 # Text generation
-npx -y bun skills/baoyu-gemini-web/scripts/main.ts "Hello"
+npx -y bun skills/baoyu-danger-gemini-web/scripts/main.ts "Hello"

 # Image generation
-npx -y bun skills/baoyu-gemini-web/scripts/main.ts --prompt "A cat" --image cat.png
+npx -y bun skills/baoyu-danger-gemini-web/scripts/main.ts --prompt "A cat" --image cat.png

 # From prompt files
-npx -y bun skills/baoyu-gemini-web/scripts/main.ts --promptfiles system.md content.md --image out.png
+npx -y bun skills/baoyu-danger-gemini-web/scripts/main.ts --promptfiles system.md content.md --image out.png
 ```

 ## Key Dependencies

 - **Bun**: TypeScript runtime (via `npx -y bun`)
- **Chrome**: Required for `baoyu-gemini-web` auth and `baoyu-post-to-x` automation
+- **Chrome**: Required for `baoyu-danger-gemini-web` auth and `baoyu-post-to-x` automation
 - **No npm packages**: Self-contained TypeScript, no external dependencies

 ## Authentication

-`baoyu-gemini-web` uses browser cookies for Google auth:
+`baoyu-danger-gemini-web` uses browser cookies for Google auth:
 - First run opens Chrome for login
 - Cookies cached in data directory
 - Force refresh: `--login` flag
@@ -70,7 +87,50 @@ npx -y bun skills/baoyu-gemini-web/scripts/main.ts --promptfiles system.md conte
   - SKILL.md `name` field: `baoyu-<name>`
 2. Add TypeScript in `skills/baoyu-<name>/scripts/`
 3. Add prompt templates in `skills/baoyu-<name>/prompts/` if needed
-4. Register in `marketplace.json` plugins[0].skills array as `./skills/baoyu-<name>`
+4. **Choose the appropriate category** and register in `marketplace.json`:
+   - `content-skills`: For content generation/publishing (images, slides, posts)
+   - `ai-generation-skills`: For AI backend capabilities
+   - `utility-skills`: For helper tools (conversion, compression)
+   - If none fit, create a new category with descriptive name
+5. **Add Script Directory section** to SKILL.md (see template below)
+
+### Choosing a Category
+
+| If your skill... | Use category |
+|------------------|--------------|
+| Generates visual content (images, slides, comics) | `content-skills` |
+| Publishes to platforms (X, WeChat, etc.) | `content-skills` |
+| Provides AI generation backend | `ai-generation-skills` |
+| Converts or processes content | `utility-skills` |
+| Compresses or optimizes files | `utility-skills` |
+
+**Creating a new category**: If the skill doesn't fit existing categories, add a new plugin object to `marketplace.json` with:
+- `name`: Descriptive kebab-case name (e.g., `analytics-skills`)
+- `description`: Brief description of the category
+- `skills`: Array with the skill path
+
+### Script Directory Template
+
+Every SKILL.md with scripts MUST include this section after Usage:
+
+```markdown
+## Script Directory
+
+**Important**: All scripts are located in the `scripts/` subdirectory of this skill.
+
+**Agent Execution Instructions**:
+1. Determine this SKILL.md file's directory path as `SKILL_DIR`
+2. Script path = `${SKILL_DIR}/scripts/<script-name>.ts`
+3. Replace all `${SKILL_DIR}` in this document with the actual path
+
+**Script Reference**:
+| Script | Purpose |
+|--------|---------|
+| `scripts/main.ts` | Main entry point |
+| `scripts/other.ts` | Other functionality |
+```
+
+When referencing scripts in workflow sections, use `${SKILL_DIR}/scripts/<name>.ts` so agents can resolve the correct path.

 ## Code Style

@@ -78,3 +138,113 @@ npx -y bun skills/baoyu-gemini-web/scripts/main.ts --promptfiles system.md conte
 - Async/await patterns
 - Short variable names
 - Type-safe interfaces
+
+## Image Generation Guidelines
+
+Skills that require image generation MUST delegate to available image generation skills rather than implementing their own.
+
+### Image Generation Skill Selection
+
+1. Check available image generation skills in `skills/` directory
+2. Read each skill's SKILL.md to understand parameters and capabilities
+3. If multiple image generation skills available, ask user to choose preferred skill
+
+### Generation Flow Template
+
+Use this template when implementing image generation in skills:
+
+```markdown
+### Step N: Generate Images
+
+**Skill Selection**:
+1. Check available image generation skills (e.g., `baoyu-danger-gemini-web`)
+2. Read selected skill's SKILL.md for parameter reference
+3. If multiple skills available, ask user to choose
+
+**Generation Flow**:
+1. Call selected image generation skill with:
+   - Prompt file path (or inline prompt)
+   - Output image path
+   - Any skill-specific parameters (refer to skill's SKILL.md)
+2. Generate images sequentially (one at a time)
+3. After each image, output progress: "Generated X/N"
+4. On failure, auto-retry once before reporting error
+```
+
+### Output Path Convention
+
+Each session creates an independent directory. Even the same source file generates a new directory per session.
+
+**Output Directory**:
+```
+<skill-suffix>/<topic-slug>/
+```
+- `<skill-suffix>`: Skill name suffix (e.g., `xhs-images`, `cover-image`, `slide-deck`, `comic`)
+- `<topic-slug>`: Generated from content topic (2-4 words, kebab-case)
+- Example: `xhs-images/ai-future-trends/`
+
+**Slug Generation**:
+1. Extract main topic from content (2-4 words, kebab-case)
+2. Example: "Introduction to Machine Learning" → `intro-machine-learning`
+
+**Conflict Resolution**:
+If `<skill-suffix>/<topic-slug>/` already exists:
+- Append timestamp: `<topic-slug>-YYYYMMDD-HHMMSS`
+- Example: `ai-future` exists → `ai-future-20260118-143052`
+
+**Source Files**:
+- Copy all sources to `<skill-suffix>/<topic-slug>/` with naming: `source-{slug}.{ext}`
+- Multiple sources supported: text, images, files from conversation
+- Examples:
+  - `source-article.md` (main text content)
+  - `source-reference.png` (image from conversation)
+  - `source-data.csv` (additional file)
+- Original source files remain unchanged
+
+### Image Naming Convention
+
+Image filenames MUST include meaningful slugs for readability:
+
+**Format**: `NN-{type}-[slug].png`
+- `NN`: Two-digit sequence number (01, 02, ...)
+- `{type}`: Image type (cover, content, page, slide, illustration, etc.)
+- `[slug]`: Descriptive kebab-case slug derived from content
+
+**Examples**:
+```
+01-cover-ai-future.png
+02-content-key-benefits.png
+03-page-enigma-machine.png
+04-slide-architecture-overview.png
+```
+
+**Slug Rules**:
+- Derived from image purpose or content (kebab-case)
+- Must be unique within the output directory
+- 2-5 words, concise but descriptive
+- When content changes significantly, update slug accordingly
+
+### Best Practices
+
+- Always read the image generation skill's SKILL.md before calling
+- Pass parameters exactly as documented in the skill
+- Handle failures gracefully with retry logic
+- Provide clear progress feedback to user
+
+## Extension Support
+
+Every SKILL.md MUST include an Extension Support section at the end:
+
+```markdown
+## Extension Support
+
+Custom styles and configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/<skill-name>/EXTEND.md` (project)
+2. `~/.baoyu-skills/<skill-name>/EXTEND.md` (user)
+
+If found, load before Step 1. Extension content overrides defaults.
+```
+
+Replace `<skill-name>` with the actual skill name (e.g., `baoyu-cover-image`).
@@ -31,13 +31,16 @@ Run the following command in Claude Code:

 1. Select **Browse and install plugins**
 2. Select **baoyu-skills**
-3. Select **content-skills**
+3. Select the plugin(s) you want to install
 4. Select **Install now**

 **Option 2: Direct Install**

 ```bash
+# Install specific plugin
 /plugin install content-skills@baoyu-skills
+/plugin install ai-generation-skills@baoyu-skills
+/plugin install utility-skills@baoyu-skills
 ```

 **Option 3: Ask the Agent**
@@ -46,6 +49,14 @@ Simply tell Claude Code:

 > Please install Skills from github.com/JimLiu/baoyu-skills

+### Available Plugins
+
+| Plugin | Description | Skills |
+|--------|-------------|--------|
+| **content-skills** | Content generation and publishing | [xhs-images](#baoyu-xhs-images), [cover-image](#baoyu-cover-image), [slide-deck](#baoyu-slide-deck), [comic](#baoyu-comic), [article-illustrator](#baoyu-article-illustrator), [post-to-x](#baoyu-post-to-x), [post-to-wechat](#baoyu-post-to-wechat) |
+| **ai-generation-skills** | AI-powered generation backends | [danger-gemini-web](#baoyu-danger-gemini-web) |
+| **utility-skills** | Utility tools for content processing | [danger-x-to-markdown](#baoyu-danger-x-to-markdown), [compress-image](#baoyu-compress-image) |
+
 ## Update Skills

 To update skills to the latest version:
@@ -61,25 +72,13 @@ You can also **Enable auto-update** to get the latest versions automatically.

 ## Available Skills

-### baoyu-gemini-web
+Skills are organized into three categories:

-Interacts with Gemini Web to generate text and images.
+### Content Skills

-**Text Generation:**
+Content generation and publishing skills.

-```bash
-/baoyu-gemini-web "Hello, Gemini"
-/baoyu-gemini-web --prompt "Explain quantum computing"
-```
-
-**Image Generation:**
-
-```bash
-/baoyu-gemini-web --prompt "A cute cat" --image cat.png
-/baoyu-gemini-web --promptfiles system.md content.md --image out.png
-```
-
-### baoyu-xhs-images
+#### baoyu-xhs-images

 Xiaohongshu (RedNote) infographic series generator. Breaks down content into 1-10 cartoon-style infographics with **Style × Layout** two-dimensional system.

@@ -112,7 +111,7 @@ Xiaohongshu (RedNote) infographic series generator. Breaks down content into 1-1
 | `comparison` | 2 sides | Before/after, pros/cons |
 | `flow` | 3-6 steps | Processes, timelines |

-### baoyu-cover-image
+#### baoyu-cover-image

 Generate hand-drawn style cover images for articles with multiple style options.

@@ -128,9 +127,28 @@ Generate hand-drawn style cover images for articles with multiple style options.
 /baoyu-cover-image path/to/article.md --no-title
 ```

-Available styles: `elegant` (default), `tech`, `warm`, `bold`, `minimal`, `playful`, `nature`, `retro`
+Available styles: `elegant` (default), `blueprint`, `bold`, `bold-editorial`, `chalkboard`, `dark-atmospheric`, `editorial-infographic`, `fantasy-animation`, `intuition-machine`, `minimal`, `nature`, `notion`, `pixel-art`, `playful`, `retro`, `sketch-notes`, `vector-illustration`, `vintage`, `warm`, `watercolor`

-### baoyu-slide-deck
+**Style Previews**:
+
+| | | |
+|:---:|:---:|:---:|
+| ![elegant](./screenshots/cover-image-styles/elegant.webp) | ![blueprint](./screenshots/cover-image-styles/blueprint.webp) | ![bold](./screenshots/cover-image-styles/bold.webp) |
+| elegant | blueprint | bold |
+| ![bold-editorial](./screenshots/cover-image-styles/bold-editorial.webp) | ![chalkboard](./screenshots/cover-image-styles/chalkboard.webp) | ![dark-atmospheric](./screenshots/cover-image-styles/dark-atmospheric.webp) |
+| bold-editorial | chalkboard | dark-atmospheric |
+| ![editorial-infographic](./screenshots/cover-image-styles/editorial-infographic.webp) | ![fantasy-animation](./screenshots/cover-image-styles/fantasy-animation.webp) | ![intuition-machine](./screenshots/cover-image-styles/intuition-machine.webp) |
+| editorial-infographic | fantasy-animation | intuition-machine |
+| ![minimal](./screenshots/cover-image-styles/minimal.webp) | ![nature](./screenshots/cover-image-styles/nature.webp) | ![notion](./screenshots/cover-image-styles/notion.webp) |
+| minimal | nature | notion |
+| ![pixel-art](./screenshots/cover-image-styles/pixel-art.webp) | ![playful](./screenshots/cover-image-styles/playful.webp) | ![retro](./screenshots/cover-image-styles/retro.webp) |
+| pixel-art | playful | retro |
+| ![sketch-notes](./screenshots/cover-image-styles/sketch-notes.webp) | ![vector-illustration](./screenshots/cover-image-styles/vector-illustration.webp) | ![vintage](./screenshots/cover-image-styles/vintage.webp) |
+| sketch-notes | vector-illustration | vintage |
+| ![warm](./screenshots/cover-image-styles/warm.webp) | ![watercolor](./screenshots/cover-image-styles/watercolor.webp) | |
+| warm | watercolor | |
+
+#### baoyu-slide-deck

 Generate professional slide deck images from content. Creates comprehensive outlines with style instructions, then generates individual slide images.

@@ -149,9 +167,46 @@ Generate professional slide deck images from content. Creates comprehensive outl
 /baoyu-slide-deck path/to/article.md --lang zh
 ```

-Available styles: `editorial` (default), `corporate`, `technical`, `playful`, `minimal`, `storytelling`, `warm`, `retro-flat`, `notion`
+**Styles** (visual aesthetics):

-### baoyu-comic
+| Style | Description | Best For |
+|-------|-------------|----------|
+| `blueprint` (default) | Technical schematics, grid texture, engineering precision | Architecture, system design |
+| `notion` | SaaS dashboard aesthetic, card-based layouts, clean data focus | Product demos, SaaS, B2B |
+| `bold-editorial` | High-impact magazine style, bold typography, dark backgrounds | Product launches, keynotes |
+| `corporate` | Navy/gold palette, structured layouts, professional icons | Investor decks, proposals |
+| `dark-atmospheric` | Cinematic dark mode, glowing accents, atmospheric depth | Entertainment, gaming, creative |
+| `editorial-infographic` | Magazine-style explainers, flat illustrations | Tech explainers, research |
+| `fantasy-animation` | Whimsical Ghibli/Disney style, hand-drawn animation | Educational, storytelling |
+| `intuition-machine` | Technical briefing, bilingual labels, aged paper texture | Technical docs, bilingual |
+| `minimal` | Ultra-clean, maximum whitespace, single accent color | Executive briefings, premium |
+| `pixel-art` | Retro 8-bit aesthetic, chunky pixels, nostalgic gaming | Gaming, developer talks |
+| `scientific` | Academic diagrams, biological pathways, precise labeling | Biology, chemistry, medical |
+| `sketch-notes` | Hand-drawn feel, soft brush strokes, warm background | Educational, tutorials |
+| `vector-illustration` | Flat vector, black outlines, retro soft colors | Creative proposals, explainers |
+| `vintage` | Aged-paper aesthetic, historical document styling | Historical, heritage, biography |
+| `watercolor` | Soft hand-painted textures, natural warmth | Lifestyle, wellness, travel |
+
+**Style Previews**:
+
+| | | |
+|:---:|:---:|:---:|
+| ![blueprint](./screenshots/slide-deck-styles/blueprint.webp) | ![chalkboard](./screenshots/slide-deck-styles/chalkboard.webp) | ![bold-editorial](./screenshots/slide-deck-styles/bold-editorial.webp) |
+| blueprint | chalkboard | bold-editorial |
+| ![corporate](./screenshots/slide-deck-styles/corporate.webp) | ![dark-atmospheric](./screenshots/slide-deck-styles/dark-atmospheric.webp) | ![editorial-infographic](./screenshots/slide-deck-styles/editorial-infographic.webp) |
+| corporate | dark-atmospheric | editorial-infographic |
+| ![fantasy-animation](./screenshots/slide-deck-styles/fantasy-animation.webp) | ![intuition-machine](./screenshots/slide-deck-styles/intuition-machine.webp) | ![minimal](./screenshots/slide-deck-styles/minimal.webp) |
+| fantasy-animation | intuition-machine | minimal |
+| ![notion](./screenshots/slide-deck-styles/notion.webp) | ![pixel-art](./screenshots/slide-deck-styles/pixel-art.webp) | ![scientific](./screenshots/slide-deck-styles/scientific.webp) |
+| notion | pixel-art | scientific |
+| ![sketch-notes](./screenshots/slide-deck-styles/sketch-notes.webp) | ![vector-illustration](./screenshots/slide-deck-styles/vector-illustration.webp) | ![vintage](./screenshots/slide-deck-styles/vintage.webp) |
+| sketch-notes | vector-illustration | vintage |
+| ![watercolor](./screenshots/slide-deck-styles/watercolor.webp) | | |
+| watercolor | | |
+
+After generation, slides are automatically merged into a `.pptx` file for easy sharing.
+
+#### baoyu-comic

 Knowledge comic creator supporting multiple styles (Logicomix/Ligne Claire, Ohmsha manga guide). Creates original educational comics with detailed panel layouts and sequential image generation.

@@ -163,24 +218,51 @@ Knowledge comic creator supporting multiple styles (Logicomix/Ligne Claire, Ohms
 /baoyu-comic posts/turing-story/source.md --style dramatic
 /baoyu-comic posts/turing-story/source.md --style ohmsha

-# Specify layout
+# Custom style (natural language)
+/baoyu-comic posts/turing-story/source.md --style "watercolor with soft edges"
+
+# Specify layout and aspect ratio
 /baoyu-comic posts/turing-story/source.md --layout cinematic
-/baoyu-comic posts/turing-story/source.md --layout webtoon
+/baoyu-comic posts/turing-story/source.md --aspect 16:9
+
+# Specify language
+/baoyu-comic posts/turing-story/source.md --lang zh

 # Direct content input
 /baoyu-comic "The story of Alan Turing and the birth of computer science"
 ```

+**Options**:
+| Option | Values |
+|--------|--------|
+| `--style` | `classic` (default), `dramatic`, `warm`, `sepia`, `vibrant`, `ohmsha`, `realistic`, `wuxia`, or custom description |
+| `--layout` | `standard` (default), `cinematic`, `dense`, `splash`, `mixed`, `webtoon` |
+| `--aspect` | `3:4` (default, portrait), `4:3` (landscape), `16:9` (widescreen) |
+| `--lang` | `auto` (default), `zh`, `en`, `ja`, etc. |
+
 **Styles** (visual aesthetics):
-| Style | Description | Best for |
+
+| Style | Description | Best For |
 |-------|-------------|----------|
-| `classic` | Traditional Ligne Claire, balanced | Biographies, balanced narratives |
-| `dramatic` | High contrast, intense | Breakthroughs, conflicts |
-| `warm` | Nostalgic, personal | Personal stories, mentorship |
-| `tech` | Modern, digital | Computing history, AI stories |
-| `sepia` | Historical, archival | Pre-1950s, classical science |
-| `vibrant` | Energetic, engaging | Science basics, young audience |
-| `ohmsha` | Manga guide style, visual metaphors | Tutorials, complex concepts (ML, physics) |
+| `classic` (default) | Traditional Ligne Claire with clean uniform outlines, flat colors, detailed backgrounds | Biographies, balanced narratives, educational content |
+| `dramatic` | High contrast with heavy shadows, intense expressions, angular compositions | Pivotal discoveries, conflicts, climactic scenes |
+| `warm` | Soft edges, golden tones, cozy interiors with nostalgic feel | Personal stories, childhood scenes, mentorship |
+| `sepia` | Vintage illustration style with aged paper effect, period-accurate details | Pre-1950s stories, classical science, historical figures |
+| `vibrant` | Energetic lines with weight variation, bright colors, dynamic poses | Science explanations, "aha" moments, young audience |
+| `ohmsha` | Manga guide style with visual metaphors, gadgets, student/mentor dynamic | Technical tutorials, complex concepts (ML, physics) |
+| `realistic` | Full-color realistic manga with digital painting, smooth gradients, accurate proportions | Wine, food, business, lifestyle, professional topics |
+| `wuxia` | Hong Kong martial arts style with ink brush strokes, dynamic combat, qi effects | Martial arts, wuxia/xianxia, Chinese historical fiction |
+
+**Style Previews**:
+
+| | | |
+|:---:|:---:|:---:|
+| ![classic](./screenshots/comic-styles/classic.webp) | ![dramatic](./screenshots/comic-styles/dramatic.webp) | ![warm](./screenshots/comic-styles/warm.webp) |
+| classic | dramatic | warm |
+| ![sepia](./screenshots/comic-styles/sepia.webp) | ![vibrant](./screenshots/comic-styles/vibrant.webp) | ![ohmsha](./screenshots/comic-styles/ohmsha.webp) |
+| sepia | vibrant | ohmsha |
+| ![realistic](./screenshots/comic-styles/realistic.webp) | ![wuxia](./screenshots/comic-styles/wuxia.webp) | |
+| realistic | wuxia | |

 **Layouts** (panel arrangement):
 | Layout | Panels/Page | Best for |
@@ -192,7 +274,39 @@ Knowledge comic creator supporting multiple styles (Logicomix/Ligne Claire, Ohms
 | `mixed` | 3-7 varies | Complex narratives, emotional arcs |
 | `webtoon` | 3-5 vertical | Ohmsha tutorials, mobile reading |

-### baoyu-post-to-wechat
+**Layout Previews**:
+
+| | | |
+|:---:|:---:|:---:|
+| ![standard](./screenshots/comic-layouts/standard.webp) | ![cinematic](./screenshots/comic-layouts/cinematic.webp) | ![dense](./screenshots/comic-layouts/dense.webp) |
+| standard | cinematic | dense |
+| ![splash](./screenshots/comic-layouts/splash.webp) | ![mixed](./screenshots/comic-layouts/mixed.webp) | ![webtoon](./screenshots/comic-layouts/webtoon.webp) |
+| splash | mixed | webtoon |
+
+#### baoyu-article-illustrator
+
+Smart article illustration skill. Analyzes article content and generates illustrations at positions requiring visual aids.
+
+```bash
+/baoyu-article-illustrator path/to/article.md
+```
+
+#### baoyu-post-to-x
+
+Post content and articles to X (Twitter). Supports regular posts with images and X Articles (long-form Markdown). Uses real Chrome with CDP to bypass anti-automation.
+
+```bash
+# Post with text
+/baoyu-post-to-x "Hello from Claude Code!"
+
+# Post with images
+/baoyu-post-to-x "Check this out" --image photo.png
+
+# Post X Article
+/baoyu-post-to-x --article path/to/article.md
+```
+
+#### baoyu-post-to-wechat

 Post content to WeChat Official Account (微信公众号). Two modes available:

@@ -214,9 +328,94 @@ Post content to WeChat Official Account (微信公众号). Two modes available:

 Prerequisites: Google Chrome installed. First run requires QR code login (session preserved).

+### AI Generation Skills
+
+AI-powered generation backends.
+
+#### baoyu-danger-gemini-web
+
+Interacts with Gemini Web to generate text and images.
+
+**Text Generation:**
+
+```bash
+/baoyu-danger-gemini-web "Hello, Gemini"
+/baoyu-danger-gemini-web --prompt "Explain quantum computing"
+```
+
+**Image Generation:**
+
+```bash
+/baoyu-danger-gemini-web --prompt "A cute cat" --image cat.png
+/baoyu-danger-gemini-web --promptfiles system.md content.md --image out.png
+```
+
+### Utility Skills
+
+Utility tools for content processing.
+
+#### baoyu-danger-x-to-markdown
+
+Converts X (Twitter) content to markdown format. Supports tweet threads and X Articles.
+
+```bash
+# Convert tweet to markdown
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456
+
+# Save to specific file
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456 -o output.md
+
+# JSON output
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456 --json
+```
+
+**Supported URLs:**
+- `https://x.com/<user>/status/<id>`
+- `https://twitter.com/<user>/status/<id>`
+- `https://x.com/i/article/<id>`
+
+**Authentication:** Uses environment variables (`X_AUTH_TOKEN`, `X_CT0`) or Chrome login for cookie-based auth.
+
+#### baoyu-compress-image
+
+Compress images to reduce file size while maintaining quality.
+
+```bash
+/baoyu-compress-image path/to/image.png
+/baoyu-compress-image path/to/images/ --quality 80
+```
+
+## Customization
+
+All skills support customization via `EXTEND.md` files. Create an extension file to override default styles, add custom configurations, or define your own presets.
+
+**Extension paths** (checked in priority order):
+1. `.baoyu-skills/<skill-name>/EXTEND.md` - Project-level (for team/project-specific settings)
+2. `~/.baoyu-skills/<skill-name>/EXTEND.md` - User-level (for personal preferences)
+
+**Example**: To customize `baoyu-cover-image` with your brand colors:
+
+```bash
+mkdir -p .baoyu-skills/baoyu-cover-image
+```
+
+Then create `.baoyu-skills/baoyu-cover-image/EXTEND.md`:
+
+```markdown
+## Custom Styles
+
+### brand
+- Primary color: #1a73e8
+- Secondary color: #34a853
+- Font style: Modern sans-serif
+- Always include company logo watermark
+```
+
+The extension content will be loaded before skill execution and override defaults.
+
 ## Disclaimer

-### baoyu-gemini-web
+### baoyu-danger-gemini-web

 This skill uses the Gemini Web API (reverse-engineered).

@@ -226,6 +425,17 @@ This skill uses the Gemini Web API (reverse-engineered).
 - Cookies are cached for subsequent runs
 - No guarantees on API stability or availability

+### baoyu-danger-x-to-markdown
+
+This skill uses a reverse-engineered X (Twitter) API.
+
+**Warning:** This is NOT an official API. Use at your own risk.
+
+- May break without notice if X changes their API
+- Account restrictions possible if API usage detected
+- First use requires consent acknowledgment
+- Authentication via environment variables or Chrome login
+
 ## License

 MIT
@@ -31,13 +31,16 @@ npx add-skill jimliu/baoyu-skills

 1. 选择 **Browse and install plugins**
 2. 选择 **baoyu-skills**
-3. 选择 **content-skills**
+3. 选择要安装的插件
 4. 选择 **Install now**

 **方式二：直接安装**

 ```bash
+# 安装指定插件
 /plugin install content-skills@baoyu-skills
+/plugin install ai-generation-skills@baoyu-skills
+/plugin install utility-skills@baoyu-skills
 ```

 **方式三：告诉 Agent**
@@ -46,6 +49,14 @@ npx add-skill jimliu/baoyu-skills

 > 请帮我安装 github.com/JimLiu/baoyu-skills 中的 Skills

+### 可用插件
+
+| 插件 | 说明 | 包含技能 |
+|------|------|----------|
+| **content-skills** | 内容生成和发布 | [xhs-images](#baoyu-xhs-images), [cover-image](#baoyu-cover-image), [slide-deck](#baoyu-slide-deck), [comic](#baoyu-comic), [article-illustrator](#baoyu-article-illustrator), [post-to-x](#baoyu-post-to-x), [post-to-wechat](#baoyu-post-to-wechat) |
+| **ai-generation-skills** | AI 生成后端 | [danger-gemini-web](#baoyu-danger-gemini-web) |
+| **utility-skills** | 内容处理工具 | [danger-x-to-markdown](#baoyu-danger-x-to-markdown), [compress-image](#baoyu-compress-image) |
+
 ## 更新技能

 更新技能到最新版本：
@@ -61,25 +72,13 @@ npx add-skill jimliu/baoyu-skills

 ## 可用技能

-### baoyu-gemini-web
+技能分为三大类：

-与 Gemini Web 交互，生成文本和图片。
+### 内容技能 (Content Skills)

-**文本生成：**
+内容生成和发布技能。

-```bash
-/baoyu-gemini-web "你好，Gemini"
-/baoyu-gemini-web --prompt "解释量子计算"
-```
-
-**图片生成：**
-
-```bash
-/baoyu-gemini-web --prompt "一只可爱的猫" --image cat.png
-/baoyu-gemini-web --promptfiles system.md content.md --image out.png
-```
-
-### baoyu-xhs-images
+#### baoyu-xhs-images

 小红书信息图系列生成器。将内容拆解为 1-10 张卡通风格信息图，支持 **风格 × 布局** 二维系统。

@@ -112,7 +111,7 @@ npx add-skill jimliu/baoyu-skills
 | `comparison` | 双栏 | 对比、优劣 |
 | `flow` | 3-6 步 | 流程、时间线 |

-### baoyu-cover-image
+#### baoyu-cover-image

 为文章生成手绘风格封面图，支持多种风格选项。

@@ -128,9 +127,28 @@ npx add-skill jimliu/baoyu-skills
 /baoyu-cover-image path/to/article.md --no-title
 ```

-可用风格：`elegant`（默认）、`tech`、`warm`、`bold`、`minimal`、`playful`、`nature`、`retro`
+可用风格：`elegant`（默认）、`blueprint`、`bold`、`bold-editorial`、`chalkboard`、`dark-atmospheric`、`editorial-infographic`、`fantasy-animation`、`intuition-machine`、`minimal`、`nature`、`notion`、`pixel-art`、`playful`、`retro`、`sketch-notes`、`vector-illustration`、`vintage`、`warm`、`watercolor`

-### baoyu-slide-deck
+**风格预览**：
+
+| | | |
+|:---:|:---:|:---:|
+| ![elegant](./screenshots/cover-image-styles/elegant.webp) | ![blueprint](./screenshots/cover-image-styles/blueprint.webp) | ![bold](./screenshots/cover-image-styles/bold.webp) |
+| elegant | blueprint | bold |
+| ![bold-editorial](./screenshots/cover-image-styles/bold-editorial.webp) | ![chalkboard](./screenshots/cover-image-styles/chalkboard.webp) | ![dark-atmospheric](./screenshots/cover-image-styles/dark-atmospheric.webp) |
+| bold-editorial | chalkboard | dark-atmospheric |
+| ![editorial-infographic](./screenshots/cover-image-styles/editorial-infographic.webp) | ![fantasy-animation](./screenshots/cover-image-styles/fantasy-animation.webp) | ![intuition-machine](./screenshots/cover-image-styles/intuition-machine.webp) |
+| editorial-infographic | fantasy-animation | intuition-machine |
+| ![minimal](./screenshots/cover-image-styles/minimal.webp) | ![nature](./screenshots/cover-image-styles/nature.webp) | ![notion](./screenshots/cover-image-styles/notion.webp) |
+| minimal | nature | notion |
+| ![pixel-art](./screenshots/cover-image-styles/pixel-art.webp) | ![playful](./screenshots/cover-image-styles/playful.webp) | ![retro](./screenshots/cover-image-styles/retro.webp) |
+| pixel-art | playful | retro |
+| ![sketch-notes](./screenshots/cover-image-styles/sketch-notes.webp) | ![vector-illustration](./screenshots/cover-image-styles/vector-illustration.webp) | ![vintage](./screenshots/cover-image-styles/vintage.webp) |
+| sketch-notes | vector-illustration | vintage |
+| ![warm](./screenshots/cover-image-styles/warm.webp) | ![watercolor](./screenshots/cover-image-styles/watercolor.webp) | |
+| warm | watercolor | |
+
+#### baoyu-slide-deck

 从内容生成专业的幻灯片图片。先创建包含样式说明的完整大纲，然后逐页生成幻灯片图片。

@@ -149,9 +167,46 @@ npx add-skill jimliu/baoyu-skills
 /baoyu-slide-deck path/to/article.md --lang zh
 ```

-可用风格：`editorial`（默认）、`corporate`、`technical`、`playful`、`minimal`、`storytelling`、`warm`、`retro-flat`、`notion`
+**风格**（视觉美学）：

-### baoyu-comic
+| 风格 | 描述 | 适用场景 |
+|------|------|----------|
+| `blueprint`（默认） | 技术蓝图风格，网格纹理，工程精度 | 架构设计、系统设计 |
+| `notion` | SaaS 仪表盘美学，卡片式布局，数据清晰 | 产品演示、SaaS、B2B |
+| `bold-editorial` | 杂志社论风格，粗体排版，深色背景 | 产品发布、主题演讲 |
+| `corporate` | 海军蓝/金色配色，结构化布局，专业图标 | 投资者演示、客户提案 |
+| `dark-atmospheric` | 电影级暗色调，发光效果，氛围感 | 娱乐、游戏、创意 |
+| `editorial-infographic` | 杂志风格信息图，扁平插画 | 科技解说、研究报告 |
+| `fantasy-animation` | 吉卜力/迪士尼风格，手绘动画 | 教育、故事讲述 |
+| `intuition-machine` | 技术简报，双语标签，做旧纸张纹理 | 技术文档、双语内容 |
+| `minimal` | 极简风格，大量留白，单一强调色 | 高管简报、高端品牌 |
+| `pixel-art` | 复古 8-bit 像素风，怀旧游戏感 | 游戏、开发者分享 |
+| `scientific` | 学术图表，生物通路，精确标注 | 生物、化学、医学 |
+| `sketch-notes` | 手绘风格，柔和笔触，暖白色背景 | 教育、教程、知识分享 |
+| `vector-illustration` | 扁平矢量风格，黑色轮廓线，复古柔和配色 | 创意提案、说明性内容 |
+| `vintage` | 做旧纸张美学，历史文档风格 | 历史、传记、人文 |
+| `watercolor` | 柔和手绘水彩纹理，自然温暖 | 生活方式、健康、旅行 |
+
+**风格预览**：
+
+| | | |
+|:---:|:---:|:---:|
+| ![blueprint](./screenshots/slide-deck-styles/blueprint.webp) | ![chalkboard](./screenshots/slide-deck-styles/chalkboard.webp) | ![bold-editorial](./screenshots/slide-deck-styles/bold-editorial.webp) |
+| blueprint | chalkboard | bold-editorial |
+| ![corporate](./screenshots/slide-deck-styles/corporate.webp) | ![dark-atmospheric](./screenshots/slide-deck-styles/dark-atmospheric.webp) | ![editorial-infographic](./screenshots/slide-deck-styles/editorial-infographic.webp) |
+| corporate | dark-atmospheric | editorial-infographic |
+| ![fantasy-animation](./screenshots/slide-deck-styles/fantasy-animation.webp) | ![intuition-machine](./screenshots/slide-deck-styles/intuition-machine.webp) | ![minimal](./screenshots/slide-deck-styles/minimal.webp) |
+| fantasy-animation | intuition-machine | minimal |
+| ![notion](./screenshots/slide-deck-styles/notion.webp) | ![pixel-art](./screenshots/slide-deck-styles/pixel-art.webp) | ![scientific](./screenshots/slide-deck-styles/scientific.webp) |
+| notion | pixel-art | scientific |
+| ![sketch-notes](./screenshots/slide-deck-styles/sketch-notes.webp) | ![vector-illustration](./screenshots/slide-deck-styles/vector-illustration.webp) | ![vintage](./screenshots/slide-deck-styles/vintage.webp) |
+| sketch-notes | vector-illustration | vintage |
+| ![watercolor](./screenshots/slide-deck-styles/watercolor.webp) | | |
+| watercolor | | |
+
+生成完成后，所有幻灯片会自动合并为 `.pptx` 文件，方便分享。
+
+#### baoyu-comic

 知识漫画创作器，支持多种风格（Logicomix/清线风格、欧姆社漫画教程风格）。创作带有详细分镜布局的原创教育漫画，逐页生成图片。

@@ -163,24 +218,51 @@ npx add-skill jimliu/baoyu-skills
 /baoyu-comic posts/turing-story/source.md --style dramatic
 /baoyu-comic posts/turing-story/source.md --style ohmsha

-# 指定布局
+# 自定义风格（自然语言描述）
+/baoyu-comic posts/turing-story/source.md --style "水彩风格，边缘柔和"
+
+# 指定布局和比例
 /baoyu-comic posts/turing-story/source.md --layout cinematic
-/baoyu-comic posts/turing-story/source.md --layout webtoon
+/baoyu-comic posts/turing-story/source.md --aspect 16:9
+
+# 指定语言
+/baoyu-comic posts/turing-story/source.md --lang zh

 # 直接输入内容
 /baoyu-comic "图灵的故事与计算机科学的诞生"
 ```

+**选项**：
+| 选项 | 取值 |
+|------|------|
+| `--style` | `classic`（默认）、`dramatic`、`warm`、`sepia`、`vibrant`、`ohmsha`、`realistic`、`wuxia`，或自然语言描述 |
+| `--layout` | `standard`（默认）、`cinematic`、`dense`、`splash`、`mixed`、`webtoon` |
+| `--aspect` | `3:4`（默认，竖版）、`4:3`（横版）、`16:9`（宽屏） |
+| `--lang` | `auto`（默认）、`zh`、`en`、`ja` 等 |
+
 **风格**（视觉美学）：
+
 | 风格 | 描述 | 适用场景 |
 |------|------|----------|
-| `classic` | 传统清线风格，平衡 | 传记、平衡叙事 |
-| `dramatic` | 高对比度，紧张感 | 突破时刻、冲突 |
-| `warm` | 怀旧、个人化 | 个人故事、师生情 |
-| `tech` | 现代、数字化 | 计算机史、AI 故事 |
-| `sepia` | 历史、档案感 | 1950 年前、古典科学 |
-| `vibrant` | 活力、吸引人 | 科普、青少年读者 |
-| `ohmsha` | 欧姆社漫画风格，视觉比喻 | 教程、复杂概念（机器学习、物理） |
+| `classic`（默认） | 传统清线风格，统一线条、平涂色彩、精细背景 | 传记、平衡叙事、教育内容 |
+| `dramatic` | 高对比度，重阴影、紧张表情、棱角分明的构图 | 重大发现、冲突、高潮场景 |
+| `warm` | 柔和边缘、金色调、温馨室内、怀旧感 | 个人故事、童年场景、师生情 |
+| `sepia` | 复古插画风格、做旧纸张效果、时代准确细节 | 1950 年前故事、古典科学、历史人物 |
+| `vibrant` | 富有活力的线条、明亮色彩、动感姿态 | 科学解说、"顿悟"时刻、青少年读者 |
+| `ohmsha` | 欧姆社漫画风格，视觉比喻、道具、学生/导师互动 | 技术教程、复杂概念（机器学习、物理） |
+| `realistic` | 全彩写实日漫风格，数字绘画、平滑渐变、准确人体比例 | 红酒、美食、商业、生活方式、专业话题 |
+| `wuxia` | 港漫武侠风格，水墨笔触、动态打斗、气功特效 | 武侠、仙侠、中国历史小说 |
+
+**风格预览**：
+
+| | | |
+|:---:|:---:|:---:|
+| ![classic](./screenshots/comic-styles/classic.webp) | ![dramatic](./screenshots/comic-styles/dramatic.webp) | ![warm](./screenshots/comic-styles/warm.webp) |
+| classic | dramatic | warm |
+| ![sepia](./screenshots/comic-styles/sepia.webp) | ![vibrant](./screenshots/comic-styles/vibrant.webp) | ![ohmsha](./screenshots/comic-styles/ohmsha.webp) |
+| sepia | vibrant | ohmsha |
+| ![realistic](./screenshots/comic-styles/realistic.webp) | ![wuxia](./screenshots/comic-styles/wuxia.webp) | |
+| realistic | wuxia | |

 **布局**（分镜排列）：
 | 布局 | 每页分镜数 | 适用场景 |
@@ -192,7 +274,39 @@ npx add-skill jimliu/baoyu-skills
 | `mixed` | 3-7 不等 | 复杂叙事、情感弧线 |
 | `webtoon` | 3-5 竖向 | 欧姆社教程、手机阅读 |

-### baoyu-post-to-wechat
+**布局预览**：
+
+| | | |
+|:---:|:---:|:---:|
+| ![standard](./screenshots/comic-layouts/standard.webp) | ![cinematic](./screenshots/comic-layouts/cinematic.webp) | ![dense](./screenshots/comic-layouts/dense.webp) |
+| standard | cinematic | dense |
+| ![splash](./screenshots/comic-layouts/splash.webp) | ![mixed](./screenshots/comic-layouts/mixed.webp) | ![webtoon](./screenshots/comic-layouts/webtoon.webp) |
+| splash | mixed | webtoon |
+
+#### baoyu-article-illustrator
+
+智能文章插图技能。分析文章内容，在需要视觉辅助的位置生成插图。
+
+```bash
+/baoyu-article-illustrator path/to/article.md
+```
+
+#### baoyu-post-to-x
+
+发布内容和文章到 X (Twitter)。支持带图片的普通帖子和 X 文章（长篇 Markdown）。使用真实 Chrome + CDP 绕过反自动化检测。
+
+```bash
+# 发布文字
+/baoyu-post-to-x "Hello from Claude Code!"
+
+# 发布带图片
+/baoyu-post-to-x "看看这个" --image photo.png
+
+# 发布 X 文章
+/baoyu-post-to-x --article path/to/article.md
+```
+
+#### baoyu-post-to-wechat

 发布内容到微信公众号，支持两种模式：

@@ -214,9 +328,94 @@ npx add-skill jimliu/baoyu-skills

 前置要求：已安装 Google Chrome，首次运行需扫码登录（登录状态会保存）

+### AI 生成技能 (AI Generation Skills)
+
+AI 驱动的生成后端。
+
+#### baoyu-danger-gemini-web
+
+与 Gemini Web 交互，生成文本和图片。
+
+**文本生成：**
+
+```bash
+/baoyu-danger-gemini-web "你好，Gemini"
+/baoyu-danger-gemini-web --prompt "解释量子计算"
+```
+
+**图片生成：**
+
+```bash
+/baoyu-danger-gemini-web --prompt "一只可爱的猫" --image cat.png
+/baoyu-danger-gemini-web --promptfiles system.md content.md --image out.png
+```
+
+### 工具技能 (Utility Skills)
+
+内容处理工具。
+
+#### baoyu-danger-x-to-markdown
+
+将 X (Twitter) 内容转换为 markdown 格式。支持推文串和 X 文章。
+
+```bash
+# 将推文转换为 markdown
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456
+
+# 保存到指定文件
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456 -o output.md
+
+# JSON 输出
+/baoyu-danger-x-to-markdown https://x.com/username/status/123456 --json
+```
+
+**支持的 URL：**
+- `https://x.com/<user>/status/<id>`
+- `https://twitter.com/<user>/status/<id>`
+- `https://x.com/i/article/<id>`
+
+**身份验证：** 使用环境变量（`X_AUTH_TOKEN`、`X_CT0`）或 Chrome 登录进行 cookie 认证。
+
+#### baoyu-compress-image
+
+压缩图片以减小文件大小，同时保持质量。
+
+```bash
+/baoyu-compress-image path/to/image.png
+/baoyu-compress-image path/to/images/ --quality 80
+```
+
+## 自定义扩展
+
+所有技能支持通过 `EXTEND.md` 文件自定义。创建扩展文件可覆盖默认样式、添加自定义配置或定义个人预设。
+
+**扩展路径**（按优先级检查）：
+1. `.baoyu-skills/<skill-name>/EXTEND.md` - 项目级（团队/项目特定设置）
+2. `~/.baoyu-skills/<skill-name>/EXTEND.md` - 用户级（个人偏好设置）
+
+**示例**：为 `baoyu-cover-image` 自定义品牌配色：
+
+```bash
+mkdir -p .baoyu-skills/baoyu-cover-image
+```
+
+然后创建 `.baoyu-skills/baoyu-cover-image/EXTEND.md`：
+
+```markdown
+## 自定义风格
+
+### brand
+- 主色：#1a73e8
+- 辅色：#34a853
+- 字体风格：现代无衬线
+- 始终包含公司 logo 水印
+```
+
+扩展内容会在技能执行前加载，并覆盖默认设置。
+
 ## 免责声明

-### baoyu-gemini-web
+### baoyu-danger-gemini-web

 此技能使用 Gemini Web API（逆向工程）。

@@ -226,6 +425,17 @@ npx add-skill jimliu/baoyu-skills
 - Cookies 会被缓存供后续使用
 - 不保证 API 的稳定性或可用性

+### baoyu-danger-x-to-markdown
+
+此技能使用逆向工程的 X (Twitter) API。
+
+**警告：** 这不是官方 API。使用风险自负。
+
+- 如果 X 更改其 API，可能会无预警失效
+- 如检测到 API 使用，账号可能受限
+- 首次使用需确认免责声明
+- 通过环境变量或 Chrome 登录进行身份验证
+
 ## 许可证

 MIT
@@ -102,22 +102,43 @@ When no `--style` is specified, analyze content to select the best style:

 ## File Management

-Save illustrations to `imgs/` subdirectory in the same folder as the article:
+### Output Directory
+
+Each session creates an independent directory named by content slug:

 ```
-path/to/
-├── article.md
-└── imgs/
-    ├── outline.md
-    ├── prompts/
-    │   ├── illustration-concept-a.md
-    │   ├── illustration-concept-b.md
-    │   └── ...
-    ├── illustration-concept-a.png
-    ├── illustration-concept-b.png
-    └── ...
+illustrations/{topic-slug}/
+├── source-{slug}.{ext}    # Source files (text, images, etc.)
+├── outline.md
+├── outline-{style}.md     # Style variant outlines
+├── prompts/
+│   ├── illustration-concept-a.md
+│   ├── illustration-concept-b.md
+│   └── ...
+├── illustration-concept-a.png
+├── illustration-concept-b.png
+└── ...
 ```

+**Slug Generation**:
+1. Extract main topic from content (2-4 words, kebab-case)
+2. Example: "The Future of AI" → `future-of-ai`
+
+### Conflict Resolution
+
+If `illustrations/{topic-slug}/` already exists:
+- Append timestamp: `{topic-slug}-YYYYMMDD-HHMMSS`
+- Example: `ai-future` exists → `ai-future-20260118-143052`
+
+### Source Files
+
+Copy all sources with naming `source-{slug}.{ext}`:
+- `source-article.md` (main text content)
+- `source-photo.jpg` (image from conversation)
+- `source-reference.pdf` (additional file)
+
+Multiple sources supported: text, images, files from conversation.
+
 ## Workflow

 ### Step 1: Analyze Content & Select Style
@@ -125,7 +146,11 @@ path/to/
 1. Read article content
 2. If `--style` specified, use that style
 3. Otherwise, scan for style signals and auto-select
-4. Extract key information:
+4. **Language detection**:
+   - Detect **source language** from article content
+   - Detect **user language** from conversation context
+   - Note if source_language ≠ user_language (will ask in Step 4)
+5. Extract key information:
   - Main topic and themes
   - Core messages per section
   - Abstract concepts needing visualization
@@ -173,10 +198,55 @@ path/to/
 ...
 ```

-### Step 4: Create Prompt Files
+### Step 4: Review & Confirm
+
+**Purpose**: Let user confirm all options in a single step before image generation.
+
+**IMPORTANT**: Present ALL options in a single confirmation step using AskUserQuestion. Do NOT interrupt workflow with multiple separate confirmations.
+
+1. **Generate 3 style variants**:
+   - Analyze content to select 3 most suitable styles
+   - Generate complete illustration plan for each style variant
+   - Save as `outline-{style}.md` (e.g., `outline-notion.md`, `outline-tech.md`, `outline-warm.md`)
+
+2. **Determine which questions to ask**:
+
+   | Question | When to Ask |
+   |----------|-------------|
+   | Style variant | Always (required) |
+   | Language | Only if `source_language ≠ user_language` |
+
+3. **Present options** (use AskUserQuestion with all applicable questions):
+
+   **Question 1 (Style)** - always:
+   - Style A (recommended): [style name] - [brief description]
+   - Style B: [style name] - [brief description]
+   - Style C: [style name] - [brief description]
+   - Custom: Provide custom style reference
+
+   **Question 2 (Language)** - only if source ≠ user language:
+   - [Source language] (matches article language)
+   - [User language] (your preference)
+
+   **Language handling**:
+   - If source language = user language: Just inform user (e.g., "Prompts will be in Chinese")
+   - If different: Ask which language to use for prompts
+
+4. **Apply selection**:
+   - Copy selected `outline-{style}.md` to `outline.md`
+   - If custom style provided, generate new plan with that style
+   - If different language selected, regenerate outline in that language
+   - User may edit `outline.md` directly for fine-tuning
+   - If modified, reload plan before proceeding
+
+5. **Proceed only after explicit user confirmation**
+
+### Step 5: Create Prompt Files

 Save prompts to `prompts/` directory with style-specific details.

+**All prompts are written in the user's confirmed language preference.**
+
 **Prompt Format**:

 ```markdown
@@ -199,7 +269,7 @@ Text content (if any):
 Style notes: [specific style characteristics]
 ```

-### Step 5: Generate Images
+### Step 6: Generate Images

 **Image Generation Skill Selection**:
 1. Check available image generation skills
@@ -212,12 +282,12 @@ Style notes: [specific style characteristics]
 4. On failure, auto-retry once
 5. If retry fails, log reason, continue to next

-### Step 6: Update Article
+### Step 7: Update Article

 Insert generated images at corresponding positions:

 ```markdown
-![illustration description](imgs/illustration-[slug].png)
+![illustration description]([article-name]/illustrations/illustration-[slug].png)
 ```

 **Insertion Rules**:
@@ -225,7 +295,7 @@ Insert generated images at corresponding positions:
 - Leave one blank line before and after image
 - Alt text uses concise description in article's language

-### Step 7: Output Summary
+### Step 8: Output Summary

 ```
 Article Illustration Complete!
@@ -244,6 +314,57 @@ Failed:
 - illustration-zzz.png: [failure reason]
 ```

+## Illustration Modification
+
+Support for modifying individual illustrations after initial generation.
+
+### Edit Single Illustration
+
+Regenerate a specific illustration with modified prompt:
+
+1. Identify illustration to edit (e.g., `illustration-concept-overview.png`)
+2. Update prompt in `prompts/illustration-concept-overview.md` if needed
+3. If content changes significantly, update slug in filename
+4. Regenerate image
+5. Update article if image reference changed
+
+### Add New Illustration
+
+Add a new illustration to the article:
+
+1. Identify insertion position in article
+2. Create new prompt with appropriate slug (e.g., `illustration-new-concept.md`)
+3. Generate new illustration image
+4. Update `outline.md` with new illustration entry
+5. Insert image reference in article at the specified position
+
+### Delete Illustration
+
+Remove an illustration from the article:
+
+1. Identify illustration to delete (e.g., `illustration-concept-overview.png`)
+2. Remove image file and prompt file
+3. Remove image reference from article
+4. Update `outline.md` to remove illustration entry
+
+### File Naming Convention
+
+Files use meaningful slugs for better readability:
+```
+illustration-[slug].png
+illustration-[slug].md (in prompts/)
+```
+
+Examples:
+- `illustration-concept-overview.png`
+- `illustration-workflow-diagram.png`
+- `illustration-key-benefits.png`
+
+**Slug rules**:
+- Derived from illustration purpose/content (kebab-case)
+- Must be unique within the article
+- When content changes significantly, update slug accordingly
+
 ## Style Reference Details

 ### elegant
@@ -325,4 +446,15 @@ Typography: Clean hand-drawn lettering, simple sans-serif labels
 - Maintain selected style consistency across all illustrations in one article
 - Image generation typically takes 10-30 seconds per image
 - Sensitive figures should use cartoon alternatives
- Prompt and illustration text language should match article language
+- Prompts written in user's confirmed language preference
+- Illustration text (labels, captions) should match article language
+
+## Extension Support
+
+Custom styles and configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/baoyu-article-illustrator/EXTEND.md` (project)
+2. `~/.baoyu-skills/baoyu-article-illustrator/EXTEND.md` (user)
+
+If found, load before Step 1. Extension content overrides defaults.
@@ -11,7 +11,6 @@ Create original knowledge comics with multiple visual styles.

 ```bash
 /baoyu-comic posts/turing-story/source.md
-/baoyu-comic posts/turing-story/source.md --style dramatic --layout cinematic
 /baoyu-comic  # then paste content
 ```

@@ -19,116 +18,371 @@ Create original knowledge comics with multiple visual styles.

 | Option | Values |
 |--------|--------|
-| `--style` | classic (default), dramatic, warm, tech, sepia, vibrant, ohmsha |
+| `--style` | classic (default), dramatic, warm, sepia, vibrant, ohmsha, realistic, wuxia, or custom description |
 | `--layout` | standard (default), cinematic, dense, splash, mixed, webtoon |
+| `--aspect` | 3:4 (default, portrait), 4:3 (landscape), 16:9 (widescreen) |
+| `--lang` | auto (default), zh, en, ja, etc. |

-Style × Layout can be freely combined.
+Style × Layout × Aspect can be freely combined. Custom styles can be described in natural language.
+
+**Aspect ratio is consistent across all pages in a comic.**

 ## Auto Selection

 | Content Signals | Style | Layout |
 |-----------------|-------|--------|
 | Tutorial, how-to, beginner | ohmsha | webtoon |
-| Computing, AI, programming | tech | dense |
+| Computing, AI, programming | ohmsha | dense |
 | Pre-1950, classical, ancient | sepia | cinematic |
 | Personal story, mentor | warm | standard |
 | Conflict, breakthrough | dramatic | splash |
+| Wine, food, business, lifestyle, professional | realistic | cinematic |
+| Martial arts, wuxia, xianxia, Chinese historical | wuxia | splash |
 | Biography, balanced | classic | mixed |

+## Script Directory
+
+**Important**: All scripts are located in the `scripts/` subdirectory of this skill.
+
+**Agent Execution Instructions**:
+1. Determine this SKILL.md file's directory path as `SKILL_DIR`
+2. Script path = `${SKILL_DIR}/scripts/<script-name>.ts`
+3. Replace all `${SKILL_DIR}` in this document with the actual path
+
+**Script Reference**:
+| Script | Purpose |
+|--------|---------|
+| `scripts/merge-to-pdf.ts` | Merge comic pages into PDF |
+
 ## File Structure

+Each session creates an independent directory named by content slug:
+
 ```
-[target]/
-├── outline.md
-├── characters/
-│   ├── characters.md    # Character definitions
-│   └── characters.png   # Character reference sheet
+comic/{topic-slug}/
+├── source-{slug}.{ext}            # Source files (text, images, etc.)
+├── analysis.md                    # Deep analysis results (YAML+MD)
+├── storyboard-chronological.md    # Variant A (preserved)
+├── storyboard-thematic.md         # Variant B (preserved)
+├── storyboard-character.md        # Variant C (preserved)
+├── characters-chronological/      # Variant A chars (preserved)
+│   ├── characters.md
+│   └── characters.png
+├── characters-thematic/           # Variant B chars (preserved)
+│   ├── characters.md
+│   └── characters.png
+├── characters-character/          # Variant C chars (preserved)
+│   ├── characters.md
+│   └── characters.png
+├── storyboard.md                  # Final selected
+├── characters/                    # Final selected
+│   ├── characters.md
+│   └── characters.png
 ├── prompts/
-│   ├── 00-cover.md
-│   └── XX-page.md
-├── 00-cover.png
-└── XX-page.png
+│   ├── 00-cover-[slug].md
+│   └── NN-page-[slug].md
+├── 00-cover-[slug].png
+├── NN-page-[slug].png
+└── {topic-slug}.pdf
 ```

-**Target directory**:
- With source path: `[source-dir]/comic/`
- Without source: `comic-outputs/YYYY-MM-DD/[topic-slug]/`
+**Slug Generation**:
+1. Extract main topic from content (2-4 words, kebab-case)
+2. Example: "Alan Turing Biography" → `alan-turing-bio`
+
+**Conflict Resolution**:
+If `comic/{topic-slug}/` already exists:
+- Append timestamp: `{topic-slug}-YYYYMMDD-HHMMSS`
+- Example: `turing-story` exists → `turing-story-20260118-143052`
+
+**Source Files**:
+Copy all sources with naming `source-{slug}.{ext}`:
+- `source-biography.md`, `source-portrait.jpg`, `source-timeline.png`, etc.
+- Multiple sources supported: text, images, files from conversation

 ## Workflow

-### Step 1: Analyze Content
+### Step 1: Analyze Content → `analysis.md`

-1. Read source content
-2. Select style (from `--style` or auto-detect)
-3. Select layout (from `--layout` or auto-detect per page)
-4. Determine page count:
+Read source content, save it if needed, and perform deep analysis.
+
+**Actions**:
+1. **Save source content** (if not already a file):
+   - If user provides a file path: use as-is
+   - If user pastes content: save to `source.md` in target directory
+2. Read source content
+3. **Deep analysis** following `references/analysis-framework.md`:
+   - Target audience identification
+   - Value proposition for readers
+   - Core themes and narrative potential
+   - Key figures and their story arcs
+4. Detect source language
+5. Determine recommended page count:
   - Short story: 5-8 pages
   - Medium complexity: 9-15 pages
   - Full biography: 16-25 pages
+6. Analyze content signals for style/layout recommendations
+7. **Save to `analysis.md`**

-### Step 2: Define Characters
+**analysis.md Format**:

-**Purpose**: Establish visual consistency across all pages.
+```yaml
+---
+title: "Alan Turing: Father of Computing"
+topic: Biography
+time_span: 1912-1954
+source_language: en
+user_language: zh
+aspect_ratio: "3:4"
+recommended_page_count: 12
+---

-1. Extract all characters from content (protagonist, supporting, antagonist, narrator)
-2. Create `characters/characters.md` with visual specs for each character
-3. Generate `characters/characters.png` (character reference sheet)
+## Target Audience

-**Reference**: `references/character-template.md` for detailed format and examples.
+- **Primary**: Tech enthusiasts curious about computing history
+- **Secondary**: Students learning about scientific breakthroughs
+- **Tertiary**: General readers interested in biographical stories

-### Step 3: Generate Outline
+## Value Proposition

-Create `outline.md` with:
- Metadata (title, style, layout, page count, character reference path)
- Cover design
- Each page: layout, panel breakdown, visual prompts
+What readers will gain:
+1. Understanding of how modern computing was born
+2. Emotional connection to a brilliant but tragic figure
+3. Appreciation for the human cost of innovation

-**Reference**: `references/outline-template.md` for detailed format.
+## Core Themes
+
+| Theme | Narrative Potential | Visual Opportunity |
+|-------|--------------------|--------------------|
+| Genius vs. Society | High conflict, dramatic arcs | Contrast scenes |
+| Code-breaking | Mystery, tension | Technical diagrams as art |
+| Personal tragedy | Emotional depth | Intimate, somber panels |
+
+## Key Figures & Story Arcs
+
+### Alan Turing (Protagonist)
+- **Arc**: Misunderstood genius → War hero → Tragic end
+- **Visual identity**: Disheveled academic, intense eyes
+- **Key moments**: Enigma breakthrough, arrest, final days
+
+### Christopher Morcom (Catalyst)
+- **Role**: Early friend whose death shaped Turing
+- **Visual identity**: Youthful, bright
+- **Key moments**: School friendship, sudden death
+
+## Content Signals
+
+- "biography" → classic + mixed
+- "computing history" → ohmsha + dense
+- "personal tragedy" → dramatic + splash
+
+## Recommended Approaches
+
+1. **Chronological** - follow life timeline (recommended for biography)
+2. **Thematic** - organize by contributions (good for educational focus)
+3. **Character-focused** - relationships drive narrative (good for emotional impact)
+```
+
+### Step 2: Generate 3 Storyboard Variants
+
+Create three distinct variants, each combining a narrative approach with a recommended style.
+
+| Variant | Narrative Approach | Recommended Style | Layout |
+|---------|-------------------|-------------------|--------|
+| A | Chronological | sepia | cinematic |
+| B | Thematic | ohmsha | dense |
+| C | Character-focused | warm | standard |
+
+**For each variant**:
+
+1. **Generate storyboard** (`storyboard-{approach}.md`):
+   - YAML front matter with narrative_approach, recommended_style, recommended_layout, aspect_ratio
+   - Cover design
+   - Each page: layout, panel breakdown, visual prompts
+   - **Written in user's preferred language**
+   - Reference: `references/storyboard-template.md`
+
+2. **Generate matching characters** (`characters-{approach}/`):
+   - `characters.md` - visual specs matching the recommended style (in user's preferred language)
+   - `characters.png` - character reference sheet
+   - Reference: `references/character-template.md`
+
+**All variants are preserved after selection for reference.**
+
+### Step 3: User Confirms All Options
+
+**IMPORTANT**: Present ALL options in a single confirmation step using AskUserQuestion. Do NOT interrupt workflow with multiple separate confirmations.
+
+**Determine which questions to ask**:
+
+| Question | When to Ask |
+|----------|-------------|
+| Storyboard variant | Always (required) |
+| Visual style | Always (required) |
+| Language | Only if `source_language ≠ user_language` |
+| Aspect ratio | Only if user might prefer non-default (e.g., landscape content) |
+
+**Language handling**:
+- If source language = user language: Just inform user (e.g., "Comic will be in Chinese")
+- If different: Ask which language to use
+
+**All storyboards and prompts are generated in the user's selected/preferred language.**
+
+**Aspect ratio handling**:
+- Default: 3:4 (portrait) - standard comic format
+- Offer 4:3 (landscape) if content suits it (e.g., panoramic scenes, technical diagrams)
+- Offer 16:9 (widescreen) for cinematic content
+
+**AskUserQuestion format** (example with all questions):
+
+```
+Question 1 (Storyboard): Which storyboard variant?
+- A: Chronological + sepia (Recommended)
+- B: Thematic + ohmsha
+- C: Character-focused + warm
+- Custom
+
+Question 2 (Style): Which visual style?
+- sepia (Recommended from variant)
+- classic / dramatic / warm / sepia / vibrant / ohmsha / realistic / wuxia
+- Custom description
+
+Question 3 (Language) - only if mismatch:
+- Chinese (source material language)
+- English (your preference)
+
+Question 4 (Aspect) - only if relevant:
+- 3:4 Portrait (Recommended)
+- 4:3 Landscape
+- 16:9 Widescreen
+```
+
+**After confirmation**:
+1. Copy selected storyboard → `storyboard.md`
+2. Copy selected characters → `characters/`
+3. Update YAML front matter with confirmed style, language, aspect_ratio
+4. If style differs from variant's recommended: regenerate `characters/characters.png`
+5. User may edit files directly for fine-tuning

 ### Step 4: Generate Images

-For each page (cover + pages):
-
-1. Save prompt to `prompts/XX-page.md`
-2. Call image generation skill with:
-   - Base prompt: `references/base-prompt.md`
-   - Character reference (text or image, depending on skill capability)
-   - Page prompt
-   - Output path
-
-**Image Generation Skill Selection**:
- Check available image generation skills in the environment
- Adapt parameters based on skill capabilities:
-  - If supports `--promptfiles`: pass prompt files
-  - If supports reference image: pass `characters/characters.png`
-  - If text-only: concatenate prompts into single text
- If multiple skills available, ask user preference
-
-**Session Management**:
-If the image generation skill supports `--sessionId`:
-1. Generate a unique session ID at the start (e.g., `comic-{topic-slug}-{timestamp}`)
-2. Use the same session ID for character sheet and all pages
-3. This ensures visual consistency (character appearance, style) across all generated images
+With confirmed storyboard + style + aspect ratio:

+**For each page (cover + pages)**:
+1. Save prompt to `prompts/NN-{cover|page}-[slug].md` (in user's preferred language)
+2. Generate image using confirmed style and aspect ratio
 3. Report progress after each generation

-### Step 5: Completion Report
+**Image Generation Skill Selection**:
+- Check available image generation skills
+- If multiple skills available, ask user preference
+
+**Character Reference Handling**:
+- If skill supports reference image: pass `characters/characters.png`
+- If skill does NOT support reference image: include `characters/characters.md` content in prompt
+
+**Session Management**:
+If image generation skill supports `--sessionId`:
+1. Generate unique session ID: `comic-{topic-slug}-{timestamp}`
+2. Use same session ID for all pages
+3. Ensures visual consistency across generated images
+
+### Step 5: Merge to PDF
+
+After all images generated:
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/merge-to-pdf.ts <comic-dir>
+```
+
+Creates `{topic-slug}.pdf` with all pages as full-page images.
+
+### Step 6: Completion Report

 ```
 Comic Complete!
-Title: [title] | Style: [style] | Pages: [count]
+Title: [title] | Style: [style] | Pages: [count] | Aspect: [ratio] | Language: [lang]
 Location: [path]
+✓ analysis.md
 ✓ characters.png
-✓ 00-cover.png ... XX-page.png
+✓ 00-cover-[slug].png ... NN-page-[slug].png
+✓ {topic-slug}.pdf
 ```

+## Page Modification
+
+Support for modifying individual pages after initial generation.
+
+### Edit Single Page
+
+Regenerate a specific page with modified prompt:
+
+1. Identify page to edit (e.g., `03-page-enigma-machine.png`)
+2. Update prompt in `prompts/03-page-enigma-machine.md` if needed
+3. If content changes significantly, update slug in filename
+4. Regenerate image using same session ID and aspect ratio
+5. Regenerate PDF
+
+### Add New Page
+
+Insert a new page at specified position:
+
+1. Specify insertion position (e.g., after page 3)
+2. Create new prompt with appropriate slug (e.g., `04-page-bletchley-park.md`)
+3. Generate new page image (same aspect ratio)
+4. **Renumber files**: All subsequent pages increment NN by 1
+   - `04-page-tragedy.png` → `05-page-tragedy.png`
+   - Slugs remain unchanged
+5. Update `storyboard.md` with new page entry
+6. Regenerate PDF
+
+### Delete Page
+
+Remove a page and renumber:
+
+1. Identify page to delete (e.g., `03-page-enigma-machine.png`)
+2. Remove image file and prompt file
+3. **Renumber files**: All subsequent pages decrement NN by 1
+   - `04-page-tragedy.png` → `03-page-tragedy.png`
+   - Slugs remain unchanged
+4. Update `storyboard.md` to remove page entry
+5. Regenerate PDF
+
+### File Naming Convention
+
+Files use meaningful slugs for better readability:
+```
+NN-cover-[slug].png / NN-page-[slug].png
+NN-cover-[slug].md / NN-page-[slug].md (in prompts/)
+```
+
+Examples:
+- `00-cover-turing-story.png`
+- `01-page-early-life.png`
+- `02-page-cambridge-years.png`
+- `03-page-enigma-machine.png`
+
+**Slug rules**:
+- Derived from page title/content (kebab-case)
+- Must be unique within the comic
+- When page content changes significantly, update slug accordingly
+
+**Renumbering**:
+- After add/delete, update NN prefix for affected pages
+- Slug remains unchanged unless content changes
+- Maintain sequential numbering with no gaps
+
 ## Style-Specific Guidelines

 ### Ohmsha Style (`--style ohmsha`)

 Additional requirements for educational manga:
- Default characters: Student (大雄), Mentor (哆啦A梦), Antagonist (胖虎)
- Custom: `--characters "Student:小明,Mentor:教授"`
+- **Default: Use Doraemon characters directly** - No need to create new characters
+  - 大雄 (Nobita): Student role, curious learner
+  - 哆啦A梦 (Doraemon): Mentor role, explains concepts with gadgets
+  - 胖虎 (Gian): Antagonist/challenge role, represents obstacles or misconceptions
+  - 静香 (Shizuka): Supporting role, asks clarifying questions
+- Custom characters only if explicitly requested: `--characters "Student:小明,Mentor:教授"`
 - Must use visual metaphors (gadgets, action scenes) - NO talking heads
 - Page titles: narrative style, not "Page X: Topic"

@@ -137,8 +391,19 @@ Additional requirements for educational manga:
 ## References

 Detailed templates and guidelines in `references/` directory:
+- `analysis-framework.md` - Deep content analysis for comic adaptation
 - `character-template.md` - Character definition format and examples
- `outline-template.md` - Outline structure and panel breakdown
+- `storyboard-template.md` - Storyboard structure and panel breakdown
 - `ohmsha-guide.md` - Ohmsha manga style specifics
 - `styles/` - Detailed style definitions
 - `layouts/` - Detailed layout definitions
+
+## Extension Support
+
+Custom styles and configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/baoyu-comic/EXTEND.md` (project)
+2. `~/.baoyu-skills/baoyu-comic/EXTEND.md` (user)
+
+If found, load before Step 1. Extension content overrides defaults.
@@ -0,0 +1,152 @@
+# Comic Content Analysis Framework
+
+Deep analysis framework for transforming source content into effective visual storytelling.
+
+## Purpose
+
+Before creating a comic, thoroughly analyze the source material to:
+- Identify the target audience and their needs
+- Determine what value the comic will deliver
+- Extract narrative potential for visual storytelling
+- Plan character arcs and key moments
+
+## Analysis Dimensions
+
+### 1. Core Content (Understanding "What")
+
+**Central Message**
+- What is the single most important idea readers should take away?
+- Can you express it in one sentence?
+
+**Key Concepts**
+- What are the essential concepts readers must understand?
+- How should these concepts be visualized?
+- Which concepts need simplified explanations?
+
+**Content Structure**
+- How is the source material organized?
+- What is the natural narrative arc?
+- Where are the climax and turning points?
+
+**Evidence & Examples**
+- What concrete examples, data, or stories support the main ideas?
+- Which examples translate well to visual panels?
+- What can be shown rather than told?
+
+### 2. Context & Background (Understanding "Why")
+
+**Source Origin**
+- Who created this content? What is their perspective?
+- What was the original purpose?
+- Is there bias to be aware of?
+
+**Historical/Cultural Context**
+- When and where does the story take place?
+- What background knowledge do readers need?
+- What period-specific visual elements are required?
+
+**Underlying Assumptions**
+- What does the source assume readers already know?
+- What implicit beliefs or values are present?
+- Should the comic challenge or reinforce these?
+
+### 3. Audience Analysis
+
+**Primary Audience**
+- Who will read this comic?
+- What is their existing knowledge level?
+- What are their interests and motivations?
+
+**Secondary Audiences**
+- Who else might benefit from this comic?
+- How might their needs differ?
+
+**Reader Questions**
+- What questions will readers have?
+- What misconceptions might they bring?
+- What "aha moments" can we create?
+
+### 4. Value Proposition
+
+**Knowledge Value**
+- What will readers learn?
+- What new perspectives will they gain?
+- How will this change their understanding?
+
+**Emotional Value**
+- What emotions should readers feel?
+- What connections will they make with characters?
+- What will make this memorable?
+
+**Practical Value**
+- Can readers apply what they learn?
+- What actions might this inspire?
+- What conversations might it spark?
+
+### 5. Narrative Potential
+
+**Story Arc Candidates**
+- What natural narratives exist in the content?
+- Where is the conflict or tension?
+- What transformations occur?
+
+**Character Potential**
+- Who are the key figures?
+- What are their motivations and obstacles?
+- How do they change throughout?
+
+**Visual Opportunities**
+- What scenes have strong visual potential?
+- Where can abstract concepts become concrete images?
+- What metaphors can be visualized?
+
+**Dramatic Moments**
+- What are the breakthrough/revelation moments?
+- Where are the emotional peaks?
+- What creates tension and release?
+
+### 6. Adaptation Considerations
+
+**What to Keep**
+- Essential facts and ideas
+- Key quotes or moments
+- Core emotional beats
+
+**What to Simplify**
+- Complex explanations
+- Dense technical details
+- Lengthy descriptions
+
+**What to Expand**
+- Brief mentions that deserve more attention
+- Implied emotions or relationships
+- Visual details not in source
+
+**What to Omit**
+- Tangential information
+- Redundant examples
+- Content that doesn't serve the narrative
+
+## Output Format
+
+Analysis results should be saved to `analysis.md` with:
+
+1. **YAML Front Matter**: Metadata (title, topic, time_span, languages, aspect_ratio, page_count)
+2. **Target Audience**: Primary, secondary, tertiary audiences with their needs
+3. **Value Proposition**: What readers will gain (knowledge, emotional, practical)
+4. **Core Themes**: Table with theme, narrative potential, visual opportunity
+5. **Key Figures & Story Arcs**: Character profiles with arcs, visual identity, key moments
+6. **Content Signals**: Style and layout recommendations based on content type
+7. **Recommended Approaches**: Narrative approaches ranked by suitability
+
+## Analysis Checklist
+
+Before proceeding to storyboard:
+
+- [ ] Can I state the core message in one sentence?
+- [ ] Do I know exactly who will read this comic?
+- [ ] Have I identified at least 3 ways this comic provides value?
+- [ ] Are there clear protagonists with compelling arcs?
+- [ ] Have I found at least 5 visually powerful moments?
+- [ ] Do I understand what to keep, simplify, expand, and omit?
+- [ ] Have I identified the emotional peaks and valleys?
@@ -20,4 +20,4 @@ Technical explanations, complex narratives, timelines

 ## Best Style Pairings

-tech, ohmsha, vibrant
+ohmsha, vibrant
@@ -27,4 +27,4 @@ Ohmsha-style tutorials, mobile reading, step-by-step guides

 ## Best Style Pairings

-ohmsha, tech, vibrant
+ohmsha, vibrant
@@ -1,23 +1,30 @@
-# Outline Template
+# Storyboard Template

-## Outline Document Format
+## Storyboard Document Format

 ```markdown
-# [Comic Title] - Knowledge Comic Outline
+---
+title: "[Comic Title]"
+topic: "[topic description]"
+time_span: "[e.g., 1912-1954]"
+narrative_approach: "[chronological/thematic/character-focused]"
+recommended_style: "[style name]"
+recommended_layout: "[layout name or varies]"
+aspect_ratio: "3:4"    # 3:4 (portrait), 4:3 (landscape), 16:9 (widescreen)
+language: "[zh/en/ja/etc.]"
+page_count: [N]
+generated: "YYYY-MM-DD HH:mm"
+---
+
+# [Comic Title] - Knowledge Comic Storyboard

-**Topic**: [topic description]
-**Time Span**: [e.g., 1912-1954]
-**Style**: [selected style]
-**Default Layout**: [selected layout or "varies"]
-**Page Count**: Cover + N pages
 **Character Reference**: characters/characters.png
-**Generated**: YYYY-MM-DD HH:mm

 ---

 ## Cover

-**Filename**: 00-cover.png
+**Filename**: 00-cover-[slug].png
 **Core Message**: [one-liner]

 **Visual Design**:
@@ -33,7 +40,7 @@

 ## Page 1 / N

-**Filename**: 01-page.png
+**Filename**: 01-page-[slug].png
 **Layout**: [standard/cinematic/dense/splash/mixed]
 **Narrative Layer**: [Main narrative / Narrator layer / Mixed]
 **Core Message**: [What this page conveys]
@@ -0,0 +1,66 @@
+# realistic
+
+Full-color realistic manga style with digital painting techniques
+
+## Style Guidelines
+
+### Line Work
+- Clean, precise outlines with clear contours
+- Uniform line weight for character definition
+- No excessive hatching - rely on color for depth
+- Smooth curves and realistic anatomical lines
+- Ligne Claire influence: clean but not overly simplified
+
+### Character Design
+- Realistic human proportions (7-8 head heights)
+- Anatomically accurate features and expressions
+- Detailed facial structure without exaggeration
+- Natural poses and body language
+- Consistent character appearance across all panels
+- Subtle expressions rather than manga-style exaggeration
+
+### Rendering Style
+- Full-color digital painting with rich gradients
+- Soft shadow transitions on skin and fabric
+- Realistic material textures (glass, liquid, fabric, wood)
+- Detailed hair with natural shine and volume
+- Environmental lighting affects all elements
+- NOT flat cel-shading - use smooth color blending
+
+### Background Treatment
+- Highly detailed, realistic environments
+- Accurate perspective and spatial depth
+- Atmospheric lighting (warm indoor, cool outdoor)
+- Professional settings rendered with precision
+- Props and objects with realistic textures
+
+## Color Palette
+
+- Skin: Natural warm tones (#F5D6C6, #E8C4B0)
+- Hair: Rich browns and blacks with highlights
+- Environment: Warm wood tones (#8B7355), cool stone (#9CA3AF)
+- Accent: Wine red (#722F37), gold (#D4AF37)
+- Lighting: Warm amber (#FFB347), cool blue (#B0C4DE)
+
+## Color Approach
+- Rich gradients for depth and volume
+- Realistic lighting with warm/cool contrast
+- Material-specific rendering (glass transparency, liquid reflection)
+- Subtle color temperature shifts for atmosphere
+- Professional, sophisticated palette
+
+## Quality Markers
+
+A good realistic manga page exhibits:
+- ✓ Anatomically accurate character proportions
+- ✓ Smooth color gradients (not flat fills)
+- ✓ Realistic material textures
+- ✓ Detailed, atmospheric backgrounds
+- ✓ Natural lighting with soft shadows
+- ✓ Expressive but subtle facial expressions
+- ✓ Professional, sophisticated aesthetic
+- ✓ Clean speech bubbles with clear typography
+
+## Best For
+
+Professional topics (wine, food, business), lifestyle content, adult-oriented narratives, educational guides for mature audiences, documentary-style storytelling
@@ -1,34 +0,0 @@
-# tech
-
-Modern, digital-age narratives
-
-## Style Guidelines
-
-### Line Work
- 2px, precise geometric undertones
- Clean, technical precision
- Circuit-like patterns in backgrounds
-
-### Character Design
- Contemporary clothing, focused expressions
- Modern tech accessories
- Clean, precise features
-
-### Background Treatment
- Digital elements, screens, circuit motifs
- Grid patterns
- Glowing interface elements
-
-## Color Palette
-
- Primary: Cyan (#00D4FF), deep blue (#1A365D), white
- Accents: Neon green (#00FF88), electric purple (#805AD5)
- Background: Dark gray (#1A202C), grid patterns
-
-## Mood
-
-Innovation, digital, contemporary
-
-## Best For
-
-Computing history, AI stories, modern tech
@@ -0,0 +1,54 @@
+# wuxia
+
+Hong Kong martial arts comic style (港漫武侠风格)
+
+## Style Guidelines
+
+### Line Work
+- 2-3px dynamic brush strokes with varying weight
+- Ink wash effects, traditional Chinese brush feel
+- Speed lines and impact effects for action scenes
+- Bold, confident strokes with sharp edges
+
+### Character Design
+- Realistic human proportions (7.5-8 head heights)
+- Defined musculature, dynamic warrior poses
+- Flowing hair and clothing in motion
+- Intense, determined facial expressions
+- Traditional Chinese martial arts attire (robes, armor)
+
+### Background Treatment
+- Dramatic landscapes: mountains, waterfalls, temples
+- Ink wash atmospheric effects
+- Flying debris, dust clouds during combat
+- Energy/qi effects with flowing lines
+- High contrast silhouettes
+
+## Color Palette
+
+- Primary: Deep black ink, crimson red (#8B0000), gold (#D4AF37)
+- Skin: Natural tan with dramatic shadows
+- Background: Misty grays, earth tones, ink wash gradients
+- Effects: Glowing qi energy (blue, gold, white)
+
+## Special Effects
+
+- **Combat impact**: Radiating lines, shockwaves, debris
+- **Qi/energy**: Flowing aura, glowing effects around characters
+- **Movement**: Extreme speed lines, motion blur
+- **Atmosphere**: Ink wash clouds, mist, flying leaves/petals
+
+## Quality Markers
+
+A good wuxia comic page exhibits:
+- ✓ Dynamic action poses with sense of motion
+- ✓ Ink brush aesthetic in line work
+- ✓ High contrast dramatic lighting
+- ✓ Atmospheric backgrounds with Chinese elements
+- ✓ Energy/qi effects that feel powerful
+- ✓ Flowing fabric and hair movement
+- ✓ Impactful combat moments
+
+## Best For
+
+Martial arts stories, Chinese historical fiction, action-heavy narratives, wuxia/xianxia adaptations
@@ -0,0 +1,116 @@
+import { existsSync, readdirSync, readFileSync } from "fs";
+import { join, basename } from "path";
+import { PDFDocument } from "pdf-lib";
+
+interface PageInfo {
+  filename: string;
+  path: string;
+  index: number;
+  promptPath?: string;
+}
+
+function parseArgs(): { dir: string; output?: string } {
+  const args = process.argv.slice(2);
+  let dir = "";
+  let output: string | undefined;
+
+  for (let i = 0; i < args.length; i++) {
+    if (args[i] === "--output" || args[i] === "-o") {
+      output = args[++i];
+    } else if (!args[i].startsWith("-")) {
+      dir = args[i];
+    }
+  }
+
+  if (!dir) {
+    console.error("Usage: bun merge-to-pdf.ts <comic-dir> [--output filename.pdf]");
+    process.exit(1);
+  }
+
+  return { dir, output };
+}
+
+function findComicPages(dir: string): PageInfo[] {
+  if (!existsSync(dir)) {
+    console.error(`Directory not found: ${dir}`);
+    process.exit(1);
+  }
+
+  const files = readdirSync(dir);
+  const pagePattern = /^(\d+)-(cover|page)(-[\w-]+)?\.(png|jpg|jpeg)$/i;
+  const promptsDir = join(dir, "prompts");
+  const hasPrompts = existsSync(promptsDir);
+
+  const pages: PageInfo[] = files
+    .filter((f) => pagePattern.test(f))
+    .map((f) => {
+      const match = f.match(pagePattern);
+      const baseName = f.replace(/\.(png|jpg|jpeg)$/i, "");
+      const promptPath = hasPrompts ? join(promptsDir, `${baseName}.md`) : undefined;
+
+      return {
+        filename: f,
+        path: join(dir, f),
+        index: parseInt(match![1], 10),
+        promptPath: promptPath && existsSync(promptPath) ? promptPath : undefined,
+      };
+    })
+    .sort((a, b) => a.index - b.index);
+
+  if (pages.length === 0) {
+    console.error(`No comic pages found in: ${dir}`);
+    console.error("Expected format: 00-cover-slug.png, 01-page-slug.png, etc.");
+    process.exit(1);
+  }
+
+  return pages;
+}
+
+async function createPdf(pages: PageInfo[], outputPath: string) {
+  const pdfDoc = await PDFDocument.create();
+  pdfDoc.setAuthor("baoyu-comic");
+  pdfDoc.setSubject("Generated Comic");
+
+  for (const page of pages) {
+    const imageData = readFileSync(page.path);
+    const ext = page.filename.toLowerCase();
+    const image = ext.endsWith(".png")
+      ? await pdfDoc.embedPng(imageData)
+      : await pdfDoc.embedJpg(imageData);
+
+    const { width, height } = image;
+    const pdfPage = pdfDoc.addPage([width, height]);
+
+    pdfPage.drawImage(image, {
+      x: 0,
+      y: 0,
+      width,
+      height,
+    });
+
+    console.log(`Added: ${page.filename}${page.promptPath ? " (prompt available)" : ""}`);
+  }
+
+  const pdfBytes = await pdfDoc.save();
+  await Bun.write(outputPath, pdfBytes);
+
+  console.log(`\nCreated: ${outputPath}`);
+  console.log(`Total pages: ${pages.length}`);
+}
+
+async function main() {
+  const { dir, output } = parseArgs();
+  const pages = findComicPages(dir);
+
+  const dirName = basename(dir) === "comic" ? basename(join(dir, "..")) : basename(dir);
+  const outputPath = output || join(dir, `${dirName}.pdf`);
+
+  console.log(`Found ${pages.length} pages in: ${dir}\n`);
+
+  await createPdf(pages, outputPath);
+}
+
+main().catch((err) => {
+  console.error("Error:", err.message);
+  process.exit(1);
+});
@@ -0,0 +1,188 @@
+---
+name: baoyu-compress-image
+description: Cross-platform image compression skill. Converts images to WebP by default with PNG-to-PNG support. Uses system tools (sips, cwebp, ImageMagick) with Sharp fallback.
+---
+
+# Image Compressor
+
+Cross-platform image compression with WebP default output, PNG-to-PNG support, preferring system tools with Sharp fallback.
+
+## Script Directory
+
+**Important**: All scripts are located in the `scripts/` subdirectory of this skill.
+
+**Agent Execution Instructions**:
+1. Determine this SKILL.md file's directory path as `SKILL_DIR`
+2. Script path = `${SKILL_DIR}/scripts/<script-name>.ts`
+3. Replace all `${SKILL_DIR}` in this document with the actual path
+
+**Script Reference**:
+| Script | Purpose |
+|--------|---------|
+| `scripts/main.ts` | CLI entry point for image compression |
+
+## Quick Start
+
+```bash
+# Compress to WebP (default)
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png
+
+# Keep original format (PNG → PNG)
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png --format png
+
+# Custom quality
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png -q 75
+
+# Process directory
+npx -y bun ${SKILL_DIR}/scripts/main.ts ./images/ -r
+```
+
+## Commands
+
+### Single File Compression
+
+```bash
+# Basic (converts to WebP, replaces original)
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png
+
+# Custom output path
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png -o compressed.webp
+
+# Keep original file
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png --keep
+
+# Custom quality (0-100, default: 80)
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png -q 75
+
+# Keep original format
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png -f png
+```
+
+### Directory Processing
+
+```bash
+# Process all images in directory
+npx -y bun ${SKILL_DIR}/scripts/main.ts ./images/
+
+# Recursive processing
+npx -y bun ${SKILL_DIR}/scripts/main.ts ./images/ -r
+
+# With custom quality
+npx -y bun ${SKILL_DIR}/scripts/main.ts ./images/ -r -q 75
+```
+
+### Output Formats
+
+```bash
+# Plain text (default)
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png
+
+# JSON output
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png --json
+```
+
+## Options
+
+| Option | Short | Description | Default |
+|--------|-------|-------------|---------|
+| `<input>` | | Input file or directory | Required |
+| `--output <path>` | `-o` | Output path | Same path, new extension |
+| `--format <fmt>` | `-f` | webp, png, jpeg | webp |
+| `--quality <n>` | `-q` | Quality 0-100 | 80 |
+| `--keep` | `-k` | Keep original file | false |
+| `--recursive` | `-r` | Process directories recursively | false |
+| `--json` | | JSON output | false |
+| `--help` | `-h` | Show help | |
+
+## Compressor Selection
+
+Priority order (auto-detected):
+
+1. **sips** (macOS built-in, WebP support since macOS 11)
+2. **cwebp** (Google's official WebP tool)
+3. **ImageMagick** (`convert` command)
+4. **Sharp** (npm package, auto-installed by Bun)
+
+The skill automatically selects the best available compressor.
+
+## Output Format
+
+### Text Mode (default)
+
+```
+image.png → image.webp (245KB → 89KB, 64% reduction)
+```
+
+### JSON Mode
+
+```json
+{
+  "input": "image.png",
+  "output": "image.webp",
+  "inputSize": 250880,
+  "outputSize": 91136,
+  "ratio": 0.36,
+  "compressor": "sips"
+}
+```
+
+### Directory JSON Mode
+
+```json
+{
+  "files": [...],
+  "summary": {
+    "totalFiles": 10,
+    "totalInputSize": 2508800,
+    "totalOutputSize": 911360,
+    "ratio": 0.36,
+    "compressor": "sips"
+  }
+}
+```
+
+## Examples
+
+### Compress single image
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts photo.png
+# photo.png → photo.webp (1.2MB → 340KB, 72% reduction)
+```
+
+### Compress with custom quality
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts photo.png -q 60
+# photo.png → photo.webp (1.2MB → 280KB, 77% reduction)
+```
+
+### Keep original format
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts screenshot.png -f png --keep
+# screenshot.png → screenshot-compressed.png (500KB → 380KB, 24% reduction)
+```
+
+### Process entire directory
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts ./screenshots/ -r
+# Processed 15 files: 12.5MB → 4.2MB (66% reduction)
+```
+
+### Get JSON for scripting
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts image.png --json | jq '.ratio'
+```
+
+## Extension Support
+
+Custom configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/baoyu-compress-image/EXTEND.md` (project)
+2. `~/.baoyu-skills/baoyu-compress-image/EXTEND.md` (user)
+
+If found, load before workflow. Extension content overrides defaults.
@@ -0,0 +1,315 @@
+#!/usr/bin/env bun
+import { existsSync, statSync, readdirSync, unlinkSync, renameSync } from "fs";
+import { basename, dirname, extname, join, resolve } from "path";
+import { spawn } from "child_process";
+
+type Compressor = "sips" | "cwebp" | "imagemagick" | "sharp";
+type Format = "webp" | "png" | "jpeg";
+
+interface Options {
+  input: string;
+  output?: string;
+  format: Format;
+  quality: number;
+  keep: boolean;
+  recursive: boolean;
+  json: boolean;
+}
+
+interface Result {
+  input: string;
+  output: string;
+  inputSize: number;
+  outputSize: number;
+  ratio: number;
+  compressor: Compressor;
+}
+
+const SUPPORTED_EXTS = [".png", ".jpg", ".jpeg", ".webp", ".gif", ".tiff"];
+
+async function commandExists(cmd: string): Promise<boolean> {
+  try {
+    const proc = spawn("which", [cmd], { stdio: "pipe" });
+    return new Promise((res) => {
+      proc.on("close", (code) => res(code === 0));
+      proc.on("error", () => res(false));
+    });
+  } catch {
+    return false;
+  }
+}
+
+async function detectCompressor(format: Format): Promise<Compressor> {
+  if (format === "webp") {
+    if (await commandExists("cwebp")) return "cwebp";
+    if (await commandExists("convert")) return "imagemagick";
+    return "sharp";
+  }
+  if (process.platform === "darwin") return "sips";
+  if (await commandExists("convert")) return "imagemagick";
+  return "sharp";
+}
+
+function runCmd(cmd: string, args: string[]): Promise<{ code: number; stderr: string }> {
+  return new Promise((res) => {
+    const proc = spawn(cmd, args, { stdio: ["ignore", "ignore", "pipe"] });
+    let stderr = "";
+    proc.stderr?.on("data", (d) => (stderr += d.toString()));
+    proc.on("close", (code) => res({ code: code ?? 1, stderr }));
+    proc.on("error", (e) => res({ code: 1, stderr: e.message }));
+  });
+}
+
+async function compressWithSips(input: string, output: string, format: Format, quality: number): Promise<void> {
+  const fmt = format === "jpeg" ? "jpeg" : format;
+  const args = ["-s", "format", fmt, "-s", "formatOptions", String(quality), input, "--out", output];
+  const { code, stderr } = await runCmd("sips", args);
+  if (code !== 0) throw new Error(`sips failed: ${stderr}`);
+}
+
+async function compressWithCwebp(input: string, output: string, quality: number): Promise<void> {
+  const args = ["-q", String(quality), input, "-o", output];
+  const { code, stderr } = await runCmd("cwebp", args);
+  if (code !== 0) throw new Error(`cwebp failed: ${stderr}`);
+}
+
+async function compressWithImagemagick(input: string, output: string, quality: number): Promise<void> {
+  const args = [input, "-quality", String(quality), output];
+  const { code, stderr } = await runCmd("convert", args);
+  if (code !== 0) throw new Error(`convert failed: ${stderr}`);
+}
+
+async function compressWithSharp(input: string, output: string, format: Format, quality: number): Promise<void> {
+  const sharp = (await import("sharp")).default;
+  let pipeline = sharp(input);
+  if (format === "webp") pipeline = pipeline.webp({ quality });
+  else if (format === "png") pipeline = pipeline.png({ quality });
+  else if (format === "jpeg") pipeline = pipeline.jpeg({ quality });
+  await pipeline.toFile(output);
+}
+
+async function compress(
+  compressor: Compressor,
+  input: string,
+  output: string,
+  format: Format,
+  quality: number
+): Promise<void> {
+  switch (compressor) {
+    case "sips":
+      await compressWithSips(input, output, format, quality);
+      break;
+    case "cwebp":
+      if (format !== "webp") {
+        await compressWithSharp(input, output, format, quality);
+      } else {
+        await compressWithCwebp(input, output, quality);
+      }
+      break;
+    case "imagemagick":
+      await compressWithImagemagick(input, output, quality);
+      break;
+    case "sharp":
+      await compressWithSharp(input, output, format, quality);
+      break;
+  }
+}
+
+function getOutputPath(input: string, format: Format, keep: boolean, customOutput?: string): string {
+  if (customOutput) return resolve(customOutput);
+  const dir = dirname(input);
+  const base = basename(input, extname(input));
+  const ext = format === "jpeg" ? ".jpg" : `.${format}`;
+  if (keep && extname(input).toLowerCase() === ext) {
+    return join(dir, `${base}-compressed${ext}`);
+  }
+  return join(dir, `${base}${ext}`);
+}
+
+function formatSize(bytes: number): string {
+  if (bytes < 1024) return `${bytes}B`;
+  if (bytes < 1024 * 1024) return `${Math.round(bytes / 1024)}KB`;
+  return `${(bytes / (1024 * 1024)).toFixed(1)}MB`;
+}
+
+async function processFile(
+  compressor: Compressor,
+  input: string,
+  opts: Options
+): Promise<Result> {
+  const absInput = resolve(input);
+  const inputSize = statSync(absInput).size;
+  const output = getOutputPath(absInput, opts.format, opts.keep, opts.output);
+  const tempOutput = output + ".tmp";
+
+  await compress(compressor, absInput, tempOutput, opts.format, opts.quality);
+
+  const outputSize = statSync(tempOutput).size;
+
+  if (!opts.keep && absInput !== output) {
+    unlinkSync(absInput);
+  }
+  renameSync(tempOutput, output);
+
+  return {
+    input: absInput,
+    output,
+    inputSize,
+    outputSize,
+    ratio: outputSize / inputSize,
+    compressor,
+  };
+}
+
+function collectFiles(dir: string, recursive: boolean): string[] {
+  const files: string[] = [];
+  const entries = readdirSync(dir, { withFileTypes: true });
+  for (const entry of entries) {
+    const full = join(dir, entry.name);
+    if (entry.isDirectory() && recursive) {
+      files.push(...collectFiles(full, recursive));
+    } else if (entry.isFile() && SUPPORTED_EXTS.includes(extname(entry.name).toLowerCase())) {
+      files.push(full);
+    }
+  }
+  return files;
+}
+
+function printHelp() {
+  console.log(`Usage: bun main.ts <input> [options]
+
+Options:
+  -o, --output <path>   Output path
+  -f, --format <fmt>    Output format: webp, png, jpeg (default: webp)
+  -q, --quality <n>     Quality 0-100 (default: 80)
+  -k, --keep            Keep original file
+  -r, --recursive       Process directories recursively
+      --json            JSON output
+  -h, --help            Show help`);
+}
+
+function parseArgs(args: string[]): Options | null {
+  const opts: Options = {
+    input: "",
+    format: "webp",
+    quality: 80,
+    keep: false,
+    recursive: false,
+    json: false,
+  };
+
+  for (let i = 0; i < args.length; i++) {
+    const arg = args[i];
+    if (arg === "-h" || arg === "--help") {
+      printHelp();
+      process.exit(0);
+    } else if (arg === "-o" || arg === "--output") {
+      opts.output = args[++i];
+    } else if (arg === "-f" || arg === "--format") {
+      const fmt = args[++i]?.toLowerCase();
+      if (fmt === "webp" || fmt === "png" || fmt === "jpeg" || fmt === "jpg") {
+        opts.format = fmt === "jpg" ? "jpeg" : (fmt as Format);
+      } else {
+        console.error(`Invalid format: ${fmt}`);
+        return null;
+      }
+    } else if (arg === "-q" || arg === "--quality") {
+      const q = parseInt(args[++i], 10);
+      if (isNaN(q) || q < 0 || q > 100) {
+        console.error(`Invalid quality: ${args[i]}`);
+        return null;
+      }
+      opts.quality = q;
+    } else if (arg === "-k" || arg === "--keep") {
+      opts.keep = true;
+    } else if (arg === "-r" || arg === "--recursive") {
+      opts.recursive = true;
+    } else if (arg === "--json") {
+      opts.json = true;
+    } else if (!arg.startsWith("-") && !opts.input) {
+      opts.input = arg;
+    }
+  }
+
+  if (!opts.input) {
+    console.error("Error: Input file or directory required");
+    printHelp();
+    return null;
+  }
+
+  return opts;
+}
+
+async function main() {
+  const args = process.argv.slice(2);
+  const opts = parseArgs(args);
+  if (!opts) process.exit(1);
+
+  const input = resolve(opts.input);
+  if (!existsSync(input)) {
+    console.error(`Error: ${input} not found`);
+    process.exit(1);
+  }
+
+  const compressor = await detectCompressor(opts.format);
+  const isDir = statSync(input).isDirectory();
+
+  if (isDir) {
+    const files = collectFiles(input, opts.recursive);
+    if (files.length === 0) {
+      console.error("No supported images found");
+      process.exit(1);
+    }
+
+    const results: Result[] = [];
+    for (const file of files) {
+      try {
+        const r = await processFile(compressor, file, { ...opts, output: undefined });
+        results.push(r);
+        if (!opts.json) {
+          const reduction = Math.round((1 - r.ratio) * 100);
+          console.log(`${r.input} → ${r.output} (${formatSize(r.inputSize)} → ${formatSize(r.outputSize)}, ${reduction}% reduction)`);
+        }
+      } catch (e) {
+        if (!opts.json) console.error(`Error processing ${file}: ${(e as Error).message}`);
+      }
+    }
+
+    if (opts.json) {
+      const totalInput = results.reduce((s, r) => s + r.inputSize, 0);
+      const totalOutput = results.reduce((s, r) => s + r.outputSize, 0);
+      console.log(
+        JSON.stringify({
+          files: results,
+          summary: {
+            totalFiles: results.length,
+            totalInputSize: totalInput,
+            totalOutputSize: totalOutput,
+            ratio: totalInput > 0 ? totalOutput / totalInput : 0,
+            compressor,
+          },
+        }, null, 2)
+      );
+    } else {
+      const totalInput = results.reduce((s, r) => s + r.inputSize, 0);
+      const totalOutput = results.reduce((s, r) => s + r.outputSize, 0);
+      const reduction = Math.round((1 - totalOutput / totalInput) * 100);
+      console.log(`\nProcessed ${results.length} files: ${formatSize(totalInput)} → ${formatSize(totalOutput)} (${reduction}% reduction)`);
+    }
+  } else {
+    try {
+      const r = await processFile(compressor, input, opts);
+      if (opts.json) {
+        console.log(JSON.stringify(r, null, 2));
+      } else {
+        const reduction = Math.round((1 - r.ratio) * 100);
+        console.log(`${r.input} → ${r.output} (${formatSize(r.inputSize)} → ${formatSize(r.outputSize)}, ${reduction}% reduction)`);
+      }
+    } catch (e) {
+      console.error(`Error: ${(e as Error).message}`);
+      process.exit(1);
+    }
+  }
+}
+
+main();
@@ -14,9 +14,9 @@ Generate hand-drawn style cover images for articles with multiple style options.
 /baoyu-cover-image path/to/article.md

 # Specify a style
-/baoyu-cover-image path/to/article.md --style tech
+/baoyu-cover-image path/to/article.md --style blueprint
 /baoyu-cover-image path/to/article.md --style warm
-/baoyu-cover-image path/to/article.md --style bold
+/baoyu-cover-image path/to/article.md --style dark-atmospheric

 # Without title text
 /baoyu-cover-image path/to/article.md --no-title
@@ -38,6 +38,8 @@ Generate hand-drawn style cover images for articles with multiple style options.
 | Option | Description |
 |--------|-------------|
 | `--style <name>` | Specify cover style (see Style Gallery below) |
+| `--aspect <ratio>` | Aspect ratio: 2.35:1 (cinematic, default), 16:9 (widescreen), 1:1 (social) |
+| `--lang <code>` | Output language for title text (en, zh, ja, etc.) |
 | `--no-title` | Generate cover without title text (visual only) |

 ## Style Gallery
@@ -45,13 +47,25 @@ Generate hand-drawn style cover images for articles with multiple style options.
 | Style | Description |
 |-------|-------------|
 | `elegant` (Default) | Refined, sophisticated, understated |
-| `tech` | Modern, clean, futuristic |
-| `warm` | Friendly, approachable, human-centered |
+| `blueprint` | Technical schematics, engineering precision |
 | `bold` | High contrast, attention-grabbing, energetic |
+| `bold-editorial` | Magazine cover impact, dramatic typography |
+| `chalkboard` | Black chalkboard, colorful chalk drawings |
+| `dark-atmospheric` | Cinematic dark mode, glowing accents |
+| `editorial-infographic` | Magazine explainer, visual storytelling |
+| `fantasy-animation` | Ghibli/Disney inspired, whimsical charm |
+| `intuition-machine` | Technical briefing, bilingual labels |
 | `minimal` | Ultra-clean, zen-like, focused |
-| `playful` | Fun, creative, whimsical |
 | `nature` | Organic, calm, earthy |
-| `retro` | Vintage, nostalgic, classic |
+| `notion` | Clean SaaS dashboard, productivity styling |
+| `pixel-art` | Retro 8-bit, nostalgic gaming aesthetic |
+| `playful` | Fun, creative, whimsical |
+| `retro` | Halftone dots, vintage badges, classic |
+| `sketch-notes` | Hand-drawn, educational, warm |
+| `vector-illustration` | Flat vector, black outlines, retro colors |
+| `vintage` | Aged paper, historical, expedition style |
+| `warm` | Friendly, approachable, human-centered |
+| `watercolor` | Soft hand-painted, natural warmth |

 Detailed style definitions: `references/styles/<style>.md`

@@ -61,58 +75,125 @@ When no `--style` is specified, the system analyzes content to select the best s

 | Content Signals | Selected Style |
 |----------------|----------------|
-| AI, coding, tech, digital, algorithm | `tech` |
+| Architecture, system design, engineering | `blueprint` |
+| Product launch, keynote, marketing, brand | `bold-editorial` |
+| Education, classroom, tutorial, teaching | `chalkboard` |
+| Entertainment, creative, premium, cinematic | `dark-atmospheric` |
+| Technology explainer, science, research | `editorial-infographic` |
+| Storytelling, children, fantasy, magical | `fantasy-animation` |
+| Technical docs, academic, bilingual | `intuition-machine` |
 | Personal story, emotion, growth, life | `warm` |
 | Controversial, urgent, must-read, warning | `bold` |
 | Simple, zen, focus, essential | `minimal` |
-| Fun, easy, beginner, casual, tutorial | `playful` |
+| Fun, easy, beginner, casual | `playful` |
 | Nature, eco, wellness, health, organic | `nature` |
-| History, classic, vintage, old, traditional | `retro` |
+| Pop culture, 80s/90s nostalgia, badges | `retro` |
+| Product, SaaS, dashboard, productivity | `notion` |
+| Gaming, retro tech, developer, 8-bit | `pixel-art` |
+| Educational, tutorial, knowledge sharing | `sketch-notes` |
+| Creative proposals, brand, toy-like | `vector-illustration` |
+| History, exploration, heritage, biography | `vintage` |
+| Lifestyle, travel, food, personal | `watercolor` |
 | Business, professional, strategy, analysis | `elegant` |

 ## File Management

-### With Article Path
+### Output Directory

-Save to `imgs/` subdirectory in the same folder as the article:
+Each session creates an independent directory named by content slug:

 ```
-path/to/
-├── article.md
-└── imgs/
-    ├── prompts/
-    │   └── cover.md
-    └── cover.png
-```
-
-### Without Article Path
-
-Save to current working directory:
-
-```
-./
-├── cover-prompt.md
+cover-image/{topic-slug}/
+├── source-{slug}.{ext}    # Source files (text, images, etc.)
+├── prompts/
+│   └── cover.md
 └── cover.png
 ```

+**Slug Generation**:
+1. Extract main topic from content (2-4 words, kebab-case)
+2. Example: "The Future of AI" → `future-of-ai`
+
+### Conflict Resolution
+
+If `cover-image/{topic-slug}/` already exists:
+- Append timestamp: `{topic-slug}-YYYYMMDD-HHMMSS`
+- Example: `ai-future` exists → `ai-future-20260118-143052`
+
+### Source Files
+
+Copy all sources with naming `source-{slug}.{ext}`:
+- `source-article.md` (main text content)
+- `source-logo.png` (image from conversation)
+
+Multiple sources supported: text, images, files from conversation.
+
 ## Workflow

 ### Step 1: Analyze Content

-Extract key information:
- **Main topic**: What is the article about?
- **Core message**: What's the key takeaway?
- **Tone**: Serious, playful, inspiring, educational?
- **Keywords**: Identify style-signaling words
+1. **Save source content** (if not already a file):
+   - If user provides a file path: use as-is
+   - If user pastes content: save to `source.md` in target directory

-### Step 2: Select Style
+2. **Extract key information**:
+   - **Main topic**: What is the article about?
+   - **Core message**: What's the key takeaway?
+   - **Tone**: Serious, playful, inspiring, educational?
+   - **Keywords**: Identify style-signaling words

-If `--style` specified, use that style. Otherwise:
-1. Scan content for style signals (see Auto Style Selection table)
-2. Match signals to most appropriate style
-3. Default to `elegant` if no clear signals
+3. **Language detection**:
+   - Detect **source language** from content
+   - Detect **user language** from conversation context
+   - Note if source_language ≠ user_language (will ask in Step 3)

-### Step 3: Generate Cover Concept
+### Step 2: Determine Options
+
+1. **Style selection**:
+   - If `--style` specified, use that style
+   - Otherwise, scan content for style signals and auto-select 3 candidates
+   - Default to `elegant` if no clear signals
+
+2. **Aspect ratio**:
+   - If `--aspect` specified, use that ratio
+   - Otherwise, prepare options: 2.35:1 (cinematic), 16:9 (widescreen), 1:1 (social)
+
+### Step 3: Confirm Options
+
+**Purpose**: Let user confirm all options in a single step before generation.
+
+**IMPORTANT**: Present ALL options in a single confirmation step using AskUserQuestion. Do NOT interrupt workflow with multiple separate confirmations.
+
+**Determine which questions to ask**:
+
+| Question | When to Ask |
+|----------|-------------|
+| Style | Always (required) |
+| Aspect ratio | Always (offer common options) |
+| Language | Only if `source_language ≠ user_language` |
+
+**Present options** (use AskUserQuestion with all applicable questions):
+
+**Question 1 (Style)** - always:
+- Style A (recommended): [style name] - [brief description]
+- Style B: [style name] - [brief description]
+- Style C: [style name] - [brief description]
+- Custom: Provide custom style reference
+
+**Question 2 (Aspect)** - always:
+- 2.35:1 Cinematic (Recommended) - ultra-wide, dramatic
+- 16:9 Widescreen - standard video/presentation
+- 1:1 Square - social media optimized
+
+**Question 3 (Language)** - only if source ≠ user language:
+- [Source language] (matches content)
+- [User language] (your preference)
+
+**Language handling**:
+- If source language = user language: Just inform user (e.g., "Title will be in Chinese")
+- If different: Ask which language to use for title text
+
+### Step 4: Generate Cover Concept

 Create a cover image concept based on selected style:

@@ -126,21 +207,26 @@ Create a cover image concept based on selected style:
 - 1-2 symbolic elements representing the topic
 - Metaphors or analogies that fit the style

-### Step 4: Create Prompt File
+### Step 5: Create Prompt File
+
+Save prompt to `prompts/cover.md` with confirmed options.
+
+**All prompts are written in the user's confirmed language preference.**

 **Prompt Format**:

 ```markdown
 Cover theme: [topic in 2-3 words]
 Style: [selected style name]
+Aspect ratio: [confirmed aspect ratio]

 [If title included:]
-Title text: [8 characters or less, in content language]
-Subtitle: [optional, in content language]
+Title text: [8 characters or less, in confirmed language]
+Subtitle: [optional, in confirmed language]

 Visual composition:
 - Main visual: [description matching style]
- Layout: [positioning based on title inclusion]
+- Layout: [positioning based on title inclusion and aspect ratio]
 - Decorative elements: [style-appropriate elements]

 Color scheme:
@@ -154,23 +240,25 @@ Style notes: [specific style characteristics to emphasize]
 Note: No title text, pure visual illustration only.
 ```

-### Step 5: Generate Image
+### Step 6: Generate Image

 **Image Generation Skill Selection**:
 1. Check available image generation skills
 2. If multiple skills available, ask user to choose

 **Generation**:
-Call selected image generation skill with prompt file and output path.
+Call selected image generation skill with prompt file, output path, and confirmed aspect ratio.

-### Step 6: Output Summary
+### Step 7: Output Summary

 ```
 Cover Image Generated!

 Topic: [topic]
 Style: [style name]
+Aspect: [aspect ratio]
 Title: [cover title] (or "No title - visual only")
+Language: [confirmed language]
 Location: [output path]

 Preview the image to verify it matches your expectations.
@@ -183,4 +271,15 @@ Preview the image to verify it matches your expectations.
 - Visual metaphors work better than literal representations
 - Maintain style consistency throughout the cover
 - Image generation typically takes 10-30 seconds
- Title text language should match content language
+- Title text uses user's confirmed language preference
+- Aspect ratio: 2.35:1 for cinematic/dramatic, 16:9 for widescreen, 1:1 for social media
+
+## Extension Support
+
+Custom styles and configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/baoyu-cover-image/EXTEND.md` (project)
+2. `~/.baoyu-skills/baoyu-cover-image/EXTEND.md` (user)
+
+If found, load before Step 1. Extension content overrides defaults.
@@ -0,0 +1,25 @@
+# blueprint
+
+Precise technical blueprint style with engineering aesthetic
+
+## Color Palette
+
+- Primary: Engineering Blue (#2563EB), Navy Blue (#1E3A5F)
+- Background: Blueprint Off-White (#FAF8F5), subtle grid overlay
+- Accents: Amber (#F59E0B), Light Blue (#BFDBFE)
+
+## Visual Elements
+
+- Precise lines with consistent stroke weights
+- Technical schematics and clean vector graphics
+- Dimension lines and measurement indicators
+- Grid alignment for all elements
+- Geometric precision for all shapes
+
+## Typography
+
+- Clean sans-serif hand lettering, technical and authoritative
+
+## Best For
+
+Technical architecture, system design, data analysis, engineering documentation
@@ -0,0 +1,26 @@
+# bold-editorial
+
+High-impact magazine editorial with bold visual expression
+
+## Color Palette
+
+- Primary: Pure White (#FFFFFF), Electric Blue (#3B82F6), Bright Orange (#FB923C)
+- Background: Deep Black (#0A0A0A), Deep Blue (#0F172A)
+- Accents: Magenta (#EC4899), Neon Green (#22C55E), Violet (#8B5CF6)
+
+## Visual Elements
+
+- Strong typography as visual element itself
+- Geometric shapes and bold color blocks
+- Full-bleed solid color backgrounds
+- Dynamic diagonal lines and angles
+- Dramatic lighting effects on text
+- Minimal decoration, maximum impact
+
+## Typography
+
+- Bold condensed typeface, oversized headlines, all-caps, tight letter-spacing
+
+## Best For
+
+Product launches, marketing, keynote speeches, brand showcases, investor pitches
@@ -0,0 +1,26 @@
+# chalkboard
+
+Black chalkboard background with colorful chalk drawing style
+
+## Color Palette
+
+- Primary: Chalk White (#F5F5F5), Chalk Yellow (#FFE566), Chalk Pink (#FF9999)
+- Background: Chalkboard Black (#1A1A1A), Dark Green-Black (#1C2B1C)
+- Accents: Chalk Blue (#66B3FF), Chalk Green (#90EE90), Chalk Orange (#FFB366)
+
+## Visual Elements
+
+- Chalkboard texture with subtle dust and smudges
+- Hand-drawn chalk illustrations with sketchy lines
+- Chalk dust effects around text and drawings
+- Imperfect lines with visible texture
+- Doodles, arrows, underlines, stars
+- Eraser marks and chalk residue
+
+## Typography
+
+- Hand-drawn chalk lettering, slightly uneven, with chalk texture
+
+## Best For
+
+Educational content, tutorials, classroom themes, back-to-school, teaching materials
@@ -0,0 +1,25 @@
+# dark-atmospheric
+
+Dark moody aesthetic with glowing accent elements
+
+## Color Palette
+
+- Primary: Electric Purple (#8B5CF6), Cyan Blue (#06B6D4), Magenta Pink (#EC4899)
+- Background: Deep Purple-Black (#0D0D1A), Rich Navy (#1A1A2E)
+- Accents: Amber (#F59E0B), Pure White (#FFFFFF)
+
+## Visual Elements
+
+- Glowing accent elements and borders
+- Subtle gradient backgrounds
+- Atmospheric fog or particle effects
+- Neon-style highlights on key elements
+- Silhouettes with backlit edges
+
+## Typography
+
+- Elegant serif or refined sans-serif in light/white with subtle glow
+
+## Best For
+
+Entertainment, creative industries, premium brands, cinematic storytelling
@@ -0,0 +1,26 @@
+# editorial-infographic
+
+Modern magazine-style editorial with clear visual storytelling
+
+## Color Palette
+
+- Primary: Near Black (#1A1A1A), Editorial Blue (#2563EB)
+- Background: Pure White (#FFFFFF), Light Gray (#F8F9FA)
+- Accents: Coral (#F97316), Emerald (#10B981), Amber (#F59E0B)
+
+## Visual Elements
+
+- Clean flat illustrations (not photos)
+- Structured multi-section layouts
+- Callout boxes for key insights
+- Icon-based data visualization
+- Visual metaphors for abstract concepts
+- Pull quotes and highlight boxes
+
+## Typography
+
+- Bold display serif or modern sans-serif, editorial sophistication
+
+## Best For
+
+Technology explainers, science communication, research summaries, thought leadership
@@ -0,0 +1,25 @@
+# fantasy-animation
+
+Whimsical hand-drawn animation style inspired by Ghibli and Disney
+
+## Color Palette
+
+- Primary: Golden Yellow (#F4D03F), Rose Pink (#E8A0BF), Deep Forest (#2D5A3D)
+- Background: Soft Sky Blue (#E8F4FC), Warm Cream (#FFF8E7)
+- Accents: Sage Green (#87A96B), Sky Blue (#7EC8E3), Coral (#F08080)
+
+## Visual Elements
+
+- Central illustrated character (friendly, expressive)
+- Magical floating objects (books, orbs, sparkles)
+- Storybook-style environment backgrounds
+- Soft shadows and gentle highlights
+- Decorative elements: stars, sparkles, flowers, leaves
+
+## Typography
+
+- Whimsical serif or decorative hand-lettered style, organic feel
+
+## Best For
+
+Educational content, storytelling, creative workshops, fantasy/gaming content
@@ -0,0 +1,26 @@
+# intuition-machine
+
+Technical briefing style with aged paper and bilingual labels
+
+## Color Palette
+
+- Primary: Dark Maroon (#5D3A3A), Teal (#2F7373)
+- Background: Aged Cream (#F5F0E6), subtle paper texture with light creases
+- Accents: Warm Brown (#8B7355), Maroon (#722F37), Deep Charcoal (#2D2D2D)
+
+## Visual Elements
+
+- Isometric 3D technical illustrations or flat 2D diagrams
+- Bilingual callout labels (English + Chinese)
+- Faded thematic background patterns
+- Clean black outlines on all elements
+- Split or triptych layouts
+- Key quote box at bottom
+
+## Typography
+
+- Bold display font in dark maroon, ALL CAPS for titles, bilingual labels
+
+## Best For
+
+Technical explanations, academic presentations, bilingual audiences, knowledge docs
@@ -0,0 +1,25 @@
+# notion
+
+Clean SaaS dashboard aesthetic with productivity tool styling
+
+## Color Palette
+
+- Primary: Near Black (#1F1F1F), Notion Blue (#2383E2)
+- Background: Light Gray (#F7F7F5), Pure White (#FFFFFF)
+- Accents: Success Green (#0F7B6C), Alert Red (#E03E3E), Warning Yellow (#DFAB01)
+
+## Visual Elements
+
+- Card-based layouts with subtle borders
+- Clean data visualizations
+- Tag and label chips
+- Icon-based navigation hints
+- Progress bars and metric displays
+
+## Typography
+
+- System UI or Inter style, clean and functional
+
+## Best For
+
+Product demos, SaaS presentations, productivity content, B2B topics
@@ -0,0 +1,26 @@
+# pixel-art
+
+Retro 8-bit pixel art aesthetic with nostalgic gaming style
+
+## Color Palette
+
+- Primary: Dark Navy (#1A1A2E), Pixel Green (#00FF00)
+- Background: Light Blue (#87CEEB), Soft Lavender (#E6E6FA)
+- Accents: Pixel Red (#FF0000), Pixel Yellow (#FFFF00), Pixel Cyan (#00FFFF)
+
+## Visual Elements
+
+- All elements with visible pixel structure
+- Simple iconography: stars, hearts, arrows
+- Text bubbles with pixel borders
+- Progress bars with chunky segments
+- Dithering patterns for gradients
+- Limited 16-32 color palette
+
+## Typography
+
+- Pixelated bitmap font style, chunky blocky letterforms
+
+## Best For
+
+Gaming content, tech tutorials, developer talks, retro-themed topics
@@ -0,0 +1,25 @@
+# sketch-notes
+
+Soft hand-drawn illustration style with fresh minimalist aesthetic
+
+## Color Palette
+
+- Primary: Deep Charcoal (#2C3E50), Soft Orange (#F4A261)
+- Background: Warm Off-White (#FAF8F0), subtle paper grain
+- Accents: Mustard Yellow (#E9C46A), Sage Green (#87A96B), Light Blue (#7EC8E3)
+
+## Visual Elements
+
+- Connection lines with hand-drawn wavy feel
+- Conceptual abstract icons illustrating ideas
+- Color fills with hand-painted casual feel
+- Doodle-style decorative elements
+- Arrows and pointers with sketchy style
+
+## Typography
+
+- Bold hand-written marker font for headlines, casual handwriting for body
+
+## Best For
+
+Educational content, knowledge sharing, technical explanations, tutorials
@@ -1,23 +0,0 @@
-# tech
-
-Modern, clean, futuristic
-
-## Color Palette
-
- Primary: Deep blue (#1A365D), electric cyan (#00D4FF), purple (#6B46C1)
- Background: Dark gray (#1A202C), near-black (#0D1117)
- Accents: Neon green (#00FF88), bright white
-
-## Visual Elements
-
- Circuit patterns, data nodes
- Geometric grids, code snippets
- Glowing effects
-
-## Typography
-
- Monospace-style hand lettering, glowing effects
-
-## Best For
-
-AI, programming, technology, digital transformation
@@ -0,0 +1,25 @@
+# vector-illustration
+
+Flat vector illustration with black outlines and retro soft colors
+
+## Color Palette
+
+- Primary: Coral Red (#E07A5F), Mint Green (#81B29A), Mustard Yellow (#F2CC8F)
+- Background: Cream Off-White (#F5F0E6), subtle paper texture
+- Accents: Burnt Orange (#D4764A), Rock Blue (#577590), Deep Charcoal (#2D2D2D) outlines
+
+## Visual Elements
+
+- All objects have closed black outlines (coloring book style)
+- Rounded line endings, avoid sharp corners
+- Simplified geometric shapes
+- 2.5D perspective with layering
+- Decorative elements: sunbursts, pill clouds, dots, stars
+
+## Typography
+
+- Large bold retro serif for titles, clean geometric sans-serif for body
+
+## Best For
+
+Educational content, creative proposals, brand showcases, explainer content
@@ -0,0 +1,26 @@
+# vintage
+
+Vintage aged-paper aesthetic with historical document styling
+
+## Color Palette
+
+- Primary: Dark Brown (#3D2914), Forest Green (#2D5A3D), Navy Blue (#1E3A5F)
+- Background: Aged Parchment (#F5E6D3), Sepia Cream (#FFF8DC)
+- Accents: Burgundy (#722F37), Gold (#C9A227), Sepia Black (#3D3D3D)
+
+## Visual Elements
+
+- Antique maps with route lines and landmarks
+- Compass roses and nautical elements
+- Specimen drawings (flora, fauna)
+- Handwritten-style annotations
+- Rope, leather, brass decorative motifs
+- Aged texture with subtle creases
+
+## Typography
+
+- Classic serif with historical character, elegant flourishes
+
+## Best For
+
+Historical content, travel, heritage storytelling, biography, scientific discovery
@@ -0,0 +1,25 @@
+# watercolor
+
+Soft watercolor illustration style with hand-painted textures
+
+## Color Palette
+
+- Primary: Soft Coral (#F4A261), Dusty Rose (#E8A0A0)
+- Background: Warm Off-White (#FAF8F0), Soft Cream (#FFF9E6)
+- Accents: Sage Green (#87A96B), Sky Blue (#7EC8E3), Soft Lavender (#C5B4E3)
+
+## Visual Elements
+
+- Watercolor washes as backgrounds
+- Visible brush strokes and textures
+- Natural elements: leaves, flowers, organic shapes
+- Color bleeds and soft edges
+- Hand-drawn arrows and connections
+
+## Typography
+
+- Elegant handwritten or brush script, organic letterforms
+
+## Best For
+
+Lifestyle content, wellness, travel guides, personal stories, creative topics
@@ -0,0 +1,267 @@
+---
+name: baoyu-danger-gemini-web
+description: Image generation skill using Gemini Web. Generates images from text prompts via Google Gemini. Also supports text generation. Use as the image generation backend for other skills like cover-image, xhs-images, article-illustrator.
+---
+
+# Gemini Web Client
+
+Supports:
+- Text generation
+- Image generation (download + save)
+- Reference images for vision input (attach local images)
+- Multi-turn conversations via persisted `--sessionId`
+
+## Script Directory
+
+**Important**: All scripts are located in the `scripts/` subdirectory of this skill.
+
+**Agent Execution Instructions**:
+1. Determine this SKILL.md file's directory path as `SKILL_DIR`
+2. Script path = `${SKILL_DIR}/scripts/<script-name>.ts`
+3. Replace all `${SKILL_DIR}` in this document with the actual path
+
+**Script Reference**:
+| Script | Purpose |
+|--------|---------|
+| `scripts/main.ts` | CLI entry point for text/image generation |
+| `scripts/gemini-webapi/*` | TypeScript port of `gemini_webapi` (GeminiClient, types, utils) |
+
+## ⚠️ Disclaimer (REQUIRED)
+
+**Before using this skill**, the consent check MUST be performed.
+
+### Consent Check Flow
+
+**Step 1**: Check consent file
+
+```bash
+# macOS
+cat ~/Library/Application\ Support/baoyu-skills/gemini-web/consent.json 2>/dev/null
+
+# Linux
+cat ~/.local/share/baoyu-skills/gemini-web/consent.json 2>/dev/null
+
+# Windows (PowerShell)
+Get-Content "$env:APPDATA\baoyu-skills\gemini-web\consent.json" 2>$null
+```
+
+**Step 2**: If consent exists and `accepted: true` with matching `disclaimerVersion: "1.0"`:
+
+Print warning and proceed:
+```
+⚠️  Warning: Using reverse-engineered Gemini Web API (not official). Accepted on: <acceptedAt date>
+```
+
+**Step 3**: If consent file doesn't exist or `disclaimerVersion` mismatch:
+
+Display disclaimer and ask user:
+
+```
+⚠️  DISCLAIMER
+
+This tool uses a reverse-engineered Gemini Web API, NOT an official Google API.
+
+Risks:
+- May break without notice if Google changes their API
+- No official support or guarantees
+- Use at your own risk
+
+Do you accept these terms and wish to continue?
+```
+
+Use `AskUserQuestion` tool with options:
+- **Yes, I accept** - Continue and save consent
+- **No, I decline** - Exit immediately
+
+**Step 4**: On acceptance, create consent file:
+
+```bash
+# macOS
+mkdir -p ~/Library/Application\ Support/baoyu-skills/gemini-web
+cat > ~/Library/Application\ Support/baoyu-skills/gemini-web/consent.json << 'EOF'
+{
+  "version": 1,
+  "accepted": true,
+  "acceptedAt": "<ISO timestamp>",
+  "disclaimerVersion": "1.0"
+}
+EOF
+
+# Linux
+mkdir -p ~/.local/share/baoyu-skills/gemini-web
+cat > ~/.local/share/baoyu-skills/gemini-web/consent.json << 'EOF'
+{
+  "version": 1,
+  "accepted": true,
+  "acceptedAt": "<ISO timestamp>",
+  "disclaimerVersion": "1.0"
+}
+EOF
+```
+
+**Step 5**: On decline, output message and stop:
+```
+User declined the disclaimer. Exiting.
+```
+
+---
+
+## Quick start
+
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello, Gemini"
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Explain quantum computing"
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cute cat" --image cat.png
+npx -y bun ${SKILL_DIR}/scripts/main.ts --promptfiles system.md content.md --image out.png
+
+# Multi-turn conversation (agent generates unique sessionId)
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Remember this: 42" --sessionId my-unique-id-123
+npx -y bun ${SKILL_DIR}/scripts/main.ts "What number?" --sessionId my-unique-id-123
+```
+
+## Commands
+
+### Text generation
+
+```bash
+# Simple prompt (positional)
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Your prompt here"
+
+# Explicit prompt flag
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Your prompt here"
+npx -y bun ${SKILL_DIR}/scripts/main.ts -p "Your prompt here"
+
+# With model selection
+npx -y bun ${SKILL_DIR}/scripts/main.ts -p "Hello" -m gemini-2.5-pro
+
+# Pipe from stdin
+echo "Summarize this" | npx -y bun ${SKILL_DIR}/scripts/main.ts
+```
+
+### Image generation
+
+```bash
+# Generate image with default path (./generated.png)
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A sunset over mountains" --image
+
+# Generate image with custom path
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "A cute robot" --image robot.png
+
+# Shorthand
+npx -y bun ${SKILL_DIR}/scripts/main.ts "A dragon" --image=dragon.png
+```
+
+### Vision input (reference images)
+
+```bash
+# Text + image -> text
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Describe this image" --reference a.png
+
+# Text + image -> image
+npx -y bun ${SKILL_DIR}/scripts/main.ts --prompt "Generate a variation" --reference a.png --image out.png
+```
+
+### Output formats
+
+```bash
+# Plain text (default)
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello"
+
+# JSON output
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello" --json
+```
+
+## Options
+
+| Option | Description |
+|--------|-------------|
+| `--prompt <text>`, `-p` | Prompt text |
+| `--promptfiles <files...>` | Read prompt from files (concatenated in order) |
+| `--model <id>`, `-m` | Model: gemini-3-pro (default), gemini-2.5-pro, gemini-2.5-flash |
+| `--image [path]` | Generate image, save to path (default: generated.png) |
+| `--reference <files...>`, `--ref <files...>` | Reference images for vision input |
+| `--sessionId <id>` | Session ID for multi-turn conversation (agent generates unique ID) |
+| `--list-sessions` | List saved sessions (max 100, sorted by update time) |
+| `--json` | Output as JSON |
+| `--login` | Refresh cookies only, then exit |
+| `--cookie-path <path>` | Custom cookie file path |
+| `--profile-dir <path>` | Chrome profile directory |
+| `--help`, `-h` | Show help |
+
+CLI note: `scripts/main.ts` supports text generation, image generation, reference images (`--reference/--ref`), and multi-turn conversations via `--sessionId`.
+
+## Models
+
+- `gemini-3-pro` - Default, latest model
+- `gemini-2.5-pro` - Previous generation pro
+- `gemini-2.5-flash` - Fast, lightweight
+
+## Authentication
+
+First run opens Chrome to authenticate with Google. Cookies are cached for subsequent runs.
+
+```bash
+# Force cookie refresh
+npx -y bun ${SKILL_DIR}/scripts/main.ts --login
+```
+
+## Environment variables
+
+| Variable | Description |
+|----------|-------------|
+| `GEMINI_WEB_DATA_DIR` | Data directory |
+| `GEMINI_WEB_COOKIE_PATH` | Cookie file path |
+| `GEMINI_WEB_CHROME_PROFILE_DIR` | Chrome profile directory |
+| `GEMINI_WEB_CHROME_PATH` | Chrome executable path |
+
+## Examples
+
+### Generate text response
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts "What is the capital of France?"
+```
+
+### Generate image
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts "A photorealistic image of a golden retriever puppy" --image puppy.png
+```
+
+### Get JSON output for parsing
+```bash
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Hello" --json | jq '.text'
+```
+
+### Generate image from prompt files
+```bash
+# Concatenate system.md + content.md as prompt
+npx -y bun ${SKILL_DIR}/scripts/main.ts --promptfiles system.md content.md --image output.png
+```
+
+### Multi-turn conversation
+```bash
+# Start a session with unique ID (agent generates this)
+npx -y bun ${SKILL_DIR}/scripts/main.ts "You are a helpful math tutor." --sessionId task-abc123
+
+# Continue the conversation (remembers context)
+npx -y bun ${SKILL_DIR}/scripts/main.ts "What is 2+2?" --sessionId task-abc123
+npx -y bun ${SKILL_DIR}/scripts/main.ts "Now multiply that by 10" --sessionId task-abc123
+
+# List recent sessions (max 100, sorted by update time)
+npx -y bun ${SKILL_DIR}/scripts/main.ts --list-sessions
+```
+
+Session files are stored in `~/Library/Application Support/baoyu-skills/gemini-web/sessions/<id>.json` and contain:
+- `id`: Session ID
+- `metadata`: Gemini chat metadata for continuation
+- `messages`: Array of `{role, content, timestamp, error?}`
+- `createdAt`, `updatedAt`: Timestamps
+
+## Extension Support
+
+Custom configurations via EXTEND.md.
+
+**Check paths** (priority order):
+1. `.baoyu-skills/baoyu-danger-gemini-web/EXTEND.md` (project)
+2. `~/.baoyu-skills/baoyu-danger-gemini-web/EXTEND.md` (user)
+
+If found, load before workflow. Extension content overrides defaults.
@@ -0,0 +1,627 @@
+import { Endpoint, ErrorCode, Headers, Model } from './constants.js';
+import { GemMixin } from './components/gem-mixin.js';
+import {
+  APIError,
+  AuthError,
+  GeminiError,
+  ImageGenerationError,
+  ModelInvalid,
+  TemporarilyBlocked,
+  TimeoutError,
+  UsageLimitExceeded,
+} from './exceptions.js';
+import { Candidate, Gem, GeneratedImage, ModelOutput, RPCData, WebImage } from './types/index.js';
+import {
+  extract_json_from_response,
+  get_access_token,
+  get_nested_value,
+  logger,
+  parse_file_name,
+  rotate_1psidts,
+  rotate_tasks,
+  fetch_with_timeout,
+  sleep,
+  upload_file,
+  write_cookie_file,
+  resolveGeminiWebCookiePath,
+} from './utils/index.js';
+
+type InitOptions = {
+  timeout?: number;
+  auto_close?: boolean;
+  close_delay?: number;
+  auto_refresh?: boolean;
+  refresh_interval?: number;
+  verbose?: boolean;
+};
+
+type RequestKwargs = RequestInit & { timeout_ms?: number };
+
+function normalize_headers(h?: HeadersInit): Record<string, string> {
+  if (!h) return {};
+  if (Array.isArray(h)) return Object.fromEntries(h.map(([k, v]) => [k, v]));
+  if (h instanceof Headers) {
+    const out: Record<string, string> = {};
+    h.forEach((v, k) => {
+      out[k] = v;
+    });
+    return out;
+  }
+  return { ...(h as Record<string, string>) };
+}
+
+function collect_strings(root: unknown, accept: (s: string) => boolean, limit: number = 20): string[] {
+  const out: string[] = [];
+  const seen = new Set<string>();
+  const stack: unknown[] = [root];
+
+  while (stack.length > 0 && out.length < limit) {
+    const v = stack.pop();
+    if (typeof v === 'string') {
+      if (accept(v) && !seen.has(v)) {
+        seen.add(v);
+        out.push(v);
+      }
+      continue;
+    }
+
+    if (Array.isArray(v)) {
+      for (let i = 0; i < v.length; i++) stack.push(v[i]);
+      continue;
+    }
+
+    if (v && typeof v === 'object') {
+      for (const val of Object.values(v as Record<string, unknown>)) stack.push(val);
+    }
+  }
+
+  return out;
+}
+
+export class GeminiClient extends GemMixin {
+  public cookies: Record<string, string> = {};
+  public proxy: string | null = null;
+  public _running: boolean = false;
+  public access_token: string | null = null;
+  public timeout: number = 300;
+  public auto_close: boolean = false;
+  public close_delay: number = 300;
+  public auto_refresh: boolean = true;
+  public refresh_interval: number = 540;
+  public kwargs: RequestInit;
+
+  private close_timer: ReturnType<typeof setTimeout> | null = null;
+  private refresh_abort: AbortController | null = null;
+
+  constructor(
+    secure_1psid: string | null = null,
+    secure_1psidts: string | null = null,
+    proxy: string | null = null,
+    kwargs: RequestInit = {},
+  ) {
+    super();
+    this.proxy = proxy;
+    this.kwargs = kwargs;
+
+    if (secure_1psid) {
+      this.cookies['__Secure-1PSID'] = secure_1psid;
+      if (secure_1psidts) this.cookies['__Secure-1PSIDTS'] = secure_1psidts;
+    }
+  }
+
+  async init(
+    timeoutOrOpts: number | InitOptions = 300,
+    auto_close: boolean = false,
+    close_delay: number = 300,
+    auto_refresh: boolean = true,
+    refresh_interval: number = 540,
+    verbose: boolean = true,
+  ): Promise<void> {
+    const opts: InitOptions =
+      typeof timeoutOrOpts === 'object'
+        ? timeoutOrOpts
+        : { timeout: timeoutOrOpts, auto_close, close_delay, auto_refresh, refresh_interval, verbose };
+
+    const timeout = opts.timeout ?? 300;
+    const ac = opts.auto_close ?? false;
+    const cd = opts.close_delay ?? 300;
+    const ar = opts.auto_refresh ?? true;
+    const ri = opts.refresh_interval ?? 540;
+    const vb = opts.verbose ?? true;
+
+    try {
+      const [token, valid] = await get_access_token(this.cookies, this.proxy, vb);
+      this.access_token = token;
+      this.cookies = valid;
+      this._running = true;
+
+      this.timeout = timeout;
+      this.auto_close = ac;
+      this.close_delay = cd;
+      if (this.auto_close) await this.reset_close_task();
+
+      this.auto_refresh = ar;
+      this.refresh_interval = ri;
+
+      const sid = this.cookies['__Secure-1PSID'];
+      if (sid) {
+        const existing = rotate_tasks.get(sid);
+        if (existing && existing instanceof AbortController) existing.abort();
+        rotate_tasks.delete(sid);
+      }
+
+      if (this.auto_refresh && sid) {
+        const ctl = new AbortController();
+        this.refresh_abort?.abort();
+        this.refresh_abort = ctl;
+        rotate_tasks.set(sid, ctl);
+        void this.start_auto_refresh(ctl.signal);
+      }
+
+      await write_cookie_file(this.cookies, resolveGeminiWebCookiePath(), 'client').catch(() => {});
+
+      if (vb) logger.success('Gemini client initialized successfully.');
+    } catch (e) {
+      await this.close();
+      throw e;
+    }
+  }
+
+  async close(delay: number = 0): Promise<void> {
+    if (delay > 0) await sleep(delay * 1000);
+    this._running = false;
+
+    if (this.close_timer) {
+      clearTimeout(this.close_timer);
+      this.close_timer = null;
+    }
+
+    this.refresh_abort?.abort();
+    this.refresh_abort = null;
+
+    const sid = this.cookies['__Secure-1PSID'];
+    const t = sid ? rotate_tasks.get(sid) : null;
+    if (t && t instanceof AbortController) t.abort();
+    if (sid) rotate_tasks.delete(sid);
+  }
+
+  async reset_close_task(): Promise<void> {
+    if (this.close_timer) {
+      clearTimeout(this.close_timer);
+      this.close_timer = null;
+    }
+
+    this.close_timer = setTimeout(() => {
+      void this.close(0);
+    }, this.close_delay * 1000);
+    this.close_timer.unref?.();
+  }
+
+  async start_auto_refresh(signal: AbortSignal): Promise<void> {
+    while (!signal.aborted) {
+      let newTs: string | null = null;
+      try {
+        newTs = await rotate_1psidts(this.cookies, this.proxy);
+      } catch (e) {
+        if (e instanceof AuthError) {
+          logger.warning('AuthError: Failed to refresh cookies. Auto refresh task canceled.');
+          return;
+        }
+        logger.warning(`Unexpected error while refreshing cookies: ${e instanceof Error ? e.message : String(e)}`);
+      }
+
+      if (newTs) {
+        this.cookies['__Secure-1PSIDTS'] = newTs;
+        await write_cookie_file(this.cookies, resolveGeminiWebCookiePath(), 'refresh').catch(() => {});
+        logger.debug('Cookies refreshed. New __Secure-1PSIDTS applied.');
+      }
+
+      await sleep(this.refresh_interval * 1000, signal);
+    }
+  }
+
+  protected async _run<T>(fn: () => Promise<T>, retry: number): Promise<T> {
+    try {
+      if (!this._running) {
+        await this.init({
+          timeout: this.timeout,
+          auto_close: this.auto_close,
+          close_delay: this.close_delay,
+          auto_refresh: this.auto_refresh,
+          refresh_interval: this.refresh_interval,
+          verbose: false,
+        });
+
+        if (!this._running) {
+          throw new APIError('Client initialization failed.');
+        }
+      }
+
+      return await fn();
+    } catch (e) {
+      let r = retry;
+      if (e instanceof ImageGenerationError) r = Math.min(1, r);
+      if (e instanceof APIError && r > 0) {
+        await sleep(1000);
+        return await this._run(fn, r - 1);
+      }
+      throw e;
+    }
+  }
+
+  async generate_content(
+    prompt: string,
+    files: string[] | null = null,
+    model: Model | string | Record<string, unknown> = Model.UNSPECIFIED,
+    gem: Gem | string | null = null,
+    chat: ChatSession | null = null,
+    kwargs: RequestKwargs = {},
+  ): Promise<ModelOutput> {
+    return await this._run(async () => {
+      if (!prompt) throw new Error('Prompt cannot be empty.');
+
+      let mdl: Model;
+      if (typeof model === 'string') mdl = Model.from_name(model);
+      else if (model instanceof Model) mdl = model;
+      else if (model && typeof model === 'object') mdl = Model.from_dict(model);
+      else throw new TypeError(`'model' must be a Model instance, string, or dictionary; got ${typeof model}`);
+
+      const gem_id = gem instanceof Gem ? gem.id : gem;
+
+      if (this.auto_close) await this.reset_close_task();
+
+      if (!this.access_token) throw new APIError('Missing access token.');
+
+      const f = files?.length ? files : null;
+      const uploaded =
+        f &&
+        (await Promise.all(
+          f.map(async (p) => [[await upload_file(p, this.proxy)], parse_file_name(p)] as [string[], string]),
+        ));
+
+      const first = uploaded ? [prompt, 0, null, uploaded] : [prompt];
+      const inner: unknown[] = [first, null, chat ? chat.metadata : null];
+
+      if (gem_id) {
+        for (let i = 0; i < 16; i++) inner.push(null);
+        inner.push(gem_id);
+      }
+
+      const f_req = JSON.stringify([null, JSON.stringify(inner)]);
+      const body = new URLSearchParams({ at: this.access_token, 'f.req': f_req }).toString();
+
+      const h0 = { ...Headers.GEMINI, ...mdl.model_header, Cookie: Object.entries(this.cookies).map(([k, v]) => `${k}=${v}`).join('; ') };
+      const h1 = { ...h0, ...normalize_headers(kwargs.headers) };
+
+      let res: Response;
+      try {
+        const timeout_ms = typeof kwargs.timeout_ms === 'number' ? kwargs.timeout_ms : this.timeout * 1000;
+        const { timeout_ms: _t, ...rest } = kwargs;
+        res = await fetch_with_timeout(Endpoint.GENERATE, {
+          method: 'POST',
+          headers: h1,
+          body,
+          redirect: 'follow',
+          ...this.kwargs,
+          ...rest,
+          timeout_ms,
+        });
+      } catch (e) {
+        throw new TimeoutError(
+          `Generate content request timed out, please try again. If the problem persists, consider setting a higher 'timeout' value when initializing GeminiClient. (${e instanceof Error ? e.message : String(e)})`,
+        );
+      }
+
+      if (res.status !== 200) {
+        await this.close();
+        throw new APIError(`Failed to generate contents. Request failed with status code ${res.status}`);
+      }
+
+      const txt = await res.text();
+      const response_json = extract_json_from_response(txt);
+
+      let body_json: unknown[] | null = null;
+      let body_index = 0;
+
+      try {
+        if (!Array.isArray(response_json)) throw new Error('Invalid JSON');
+        for (let part_index = 0; part_index < response_json.length; part_index++) {
+          const part = response_json[part_index];
+          if (!Array.isArray(part)) continue;
+          const part_body = get_nested_value<string | null>(part, [2], null);
+          if (!part_body) continue;
+          try {
+            const part_json = JSON.parse(part_body) as unknown[];
+            if (get_nested_value(part_json, [4], null)) {
+              body_index = part_index;
+              body_json = part_json;
+              break;
+            }
+          } catch {}
+        }
+        if (!body_json) throw new Error('No body');
+      } catch {
+        await this.close();
+        try {
+          const code = get_nested_value<number>(response_json, [0, 5, 2, 0, 1, 0], -1);
+          if (code === ErrorCode.USAGE_LIMIT_EXCEEDED) {
+            throw new UsageLimitExceeded(
+              `Failed to generate contents. Usage limit of ${mdl.model_name} model has exceeded. Please try switching to another model.`,
+            );
+          }
+          if (code === ErrorCode.MODEL_INCONSISTENT) {
+            throw new ModelInvalid(
+              'Failed to generate contents. The specified model is inconsistent with the chat history. Please make sure to pass the same `model` parameter when starting a chat session with previous metadata.',
+            );
+          }
+          if (code === ErrorCode.MODEL_HEADER_INVALID) {
+            throw new ModelInvalid(
+              'Failed to generate contents. The specified model is not available. Please update gemini_webapi to the latest version. If the error persists and is caused by the package, please report it on GitHub.',
+            );
+          }
+          if (code === ErrorCode.IP_TEMPORARILY_BLOCKED) {
+            throw new TemporarilyBlocked(
+              'Failed to generate contents. Your IP address is temporarily blocked by Google. Please try using a proxy or waiting for a while.',
+            );
+          }
+        } catch (e) {
+          if (e instanceof GeminiError) throw e;
+        }
+
+        logger.debug(`Invalid response: ${txt.slice(0, 500)}`);
+        throw new APIError('Failed to generate contents. Invalid response data received. Client will try to re-initialize on next request.');
+      }
+
+      try {
+        const candidate_list = get_nested_value<unknown[]>(body_json, [4], []);
+        const out: Candidate[] = [];
+
+        for (let candidate_index = 0; candidate_index < candidate_list.length; candidate_index++) {
+          const candidate = candidate_list[candidate_index];
+          if (!Array.isArray(candidate)) continue;
+
+          const rcid = get_nested_value<string | null>(candidate, [0], null);
+          if (!rcid) continue;
+
+          let text = String(get_nested_value(candidate, [1, 0], ''));
+          if (/^http:\/\/googleusercontent\.com\/card_content\/\d+/.test(text)) {
+            text = String(get_nested_value(candidate, [22, 0], text));
+          }
+
+          const thoughts = get_nested_value<string | null>(candidate, [37, 0, 0], null);
+
+          const web_images: WebImage[] = [];
+          for (const w of get_nested_value<unknown[]>(candidate, [12, 1], [])) {
+            if (!Array.isArray(w)) continue;
+            const url = get_nested_value<string | null>(w, [0, 0, 0], null);
+            if (!url) continue;
+            web_images.push(new WebImage(url, String(get_nested_value(w, [7, 0], '')), String(get_nested_value(w, [0, 4], '')), this.proxy));
+          }
+
+          const generated_images: GeneratedImage[] = [];
+          const wants_generated =
+            get_nested_value(candidate, [12, 7, 0], null) != null ||
+            /http:\/\/googleusercontent\.com\/image_generation_content\/\d+/.test(text);
+
+          if (wants_generated) {
+            let img_body: unknown[] | null = null;
+            for (let part_index = body_index; part_index < (response_json as unknown[]).length; part_index++) {
+              const part = (response_json as unknown[])[part_index];
+              if (!Array.isArray(part)) continue;
+              const part_body = get_nested_value<string | null>(part, [2], null);
+              if (!part_body) continue;
+              try {
+                const part_json = JSON.parse(part_body) as unknown[];
+                const cand = get_nested_value<unknown>(part_json, [4, candidate_index], null);
+                if (!cand) continue;
+
+                const urls = collect_strings(cand, (s) => s.startsWith('https://lh3.googleusercontent.com/gg-dl/'), 1);
+                if (urls.length > 0) {
+                  img_body = part_json;
+                  break;
+                }
+              } catch {}
+            }
+
+            if (!img_body) {
+              throw new ImageGenerationError(
+                'Failed to parse generated images. Please update gemini_webapi to the latest version. If the error persists and is caused by the package, please report it on GitHub.',
+              );
+            }
+
+            const img_candidate = get_nested_value<unknown[]>(img_body, [4, candidate_index], []);
+            const finished = get_nested_value<string | null>(img_candidate, [1, 0], null);
+            if (finished) {
+              text = finished.replace(/http:\/\/googleusercontent\.com\/image_generation_content\/\d+/g, '').trimEnd();
+            }
+
+            const gen = get_nested_value<unknown[]>(img_candidate, [12, 7, 0], []);
+            for (let img_index = 0; img_index < gen.length; img_index++) {
+              const g = gen[img_index];
+              if (!Array.isArray(g)) continue;
+              const url = get_nested_value<string | null>(g, [0, 3, 3], null);
+              if (!url) continue;
+              const img_num = get_nested_value<number | null>(g, [3, 6], null);
+              const title = img_num ? `[Generated Image ${img_num}]` : '[Generated Image]';
+              const alt_list = get_nested_value<unknown[]>(g, [3, 5], []);
+              const alt =
+                (typeof alt_list[img_index] === 'string' ? (alt_list[img_index] as string) : null) ??
+                (typeof alt_list[0] === 'string' ? (alt_list[0] as string) : '') ??
+                '';
+              generated_images.push(new GeneratedImage(url, title, alt, this.proxy, this.cookies));
+            }
+
+            if (generated_images.length === 0) {
+              const urls = collect_strings(img_candidate, (s) => s.startsWith('https://lh3.googleusercontent.com/gg-dl/'), 4);
+              for (const url of urls) {
+                generated_images.push(new GeneratedImage(url, '[Generated Image]', '', this.proxy, this.cookies));
+              }
+            }
+          }
+
+          out.push(new Candidate({ rcid, text, thoughts, web_images, generated_images }));
+        }
+
+        if (out.length === 0) {
+          throw new GeminiError('Failed to generate contents. No output data found in response.');
+        }
+
+        const metadata = get_nested_value<string[]>(body_json, [1], []);
+        const output = new ModelOutput({ metadata, candidates: out });
+
+        if (chat instanceof ChatSession) chat.last_output = output;
+        return output;
+      } catch (e) {
+        if (e instanceof GeminiError || e instanceof APIError) throw e;
+        throw new APIError('Failed to parse response body. Data structure is invalid.');
+      }
+    }, 2);
+  }
+
+  async generateContent(
+    prompt: string,
+    files?: string[] | null,
+    model?: Model | string | Record<string, unknown>,
+    gem?: Gem | string | null,
+    chat?: ChatSession | null,
+    kwargs?: RequestKwargs,
+  ): Promise<ModelOutput> {
+    return await this.generate_content(prompt, files ?? null, model ?? Model.UNSPECIFIED, gem ?? null, chat ?? null, kwargs ?? {});
+  }
+
+  start_chat(opts?: ConstructorParameters<typeof ChatSession>[1]): ChatSession {
+    return new ChatSession(this, opts);
+  }
+
+  startChat(opts?: ConstructorParameters<typeof ChatSession>[1]): ChatSession {
+    return this.start_chat(opts);
+  }
+
+  protected async _batch_execute(payloads: RPCData[], opts: RequestInit = {}): Promise<Response> {
+    if (!this.access_token) throw new APIError('Missing access token.');
+
+    const f_req = JSON.stringify([payloads.map((p) => p.serialize())]);
+    const body = new URLSearchParams({ at: this.access_token, 'f.req': f_req }).toString();
+
+    const h0 = { ...Headers.GEMINI, Cookie: Object.entries(this.cookies).map(([k, v]) => `${k}=${v}`).join('; ') };
+    const h1 = { ...h0, ...normalize_headers(opts.headers) };
+
+    const res = await fetch_with_timeout(Endpoint.BATCH_EXEC, {
+      method: 'POST',
+      headers: h1,
+      body,
+      redirect: 'follow',
+      ...this.kwargs,
+      ...opts,
+      timeout_ms: this.timeout * 1000,
+    });
+
+    if (res.status !== 200) {
+      await this.close();
+      throw new APIError(`Batch execution failed with status code ${res.status}`);
+    }
+
+    return res;
+  }
+}
+
+export class ChatSession {
+  private __metadata: Array<string | null> = [null, null, null];
+  public geminiclient: GeminiClient;
+  private _last_output: ModelOutput | null = null;
+  public model: Model | string | Record<string, unknown>;
+  public gem: Gem | string | null;
+
+  constructor(
+    geminiclient: GeminiClient,
+    opts: {
+      metadata?: Array<string | null>;
+      cid?: string | null;
+      rid?: string | null;
+      rcid?: string | null;
+      model?: Model | string | Record<string, unknown>;
+      gem?: Gem | string | null;
+    } = {},
+  ) {
+    this.geminiclient = geminiclient;
+    this.model = opts.model ?? Model.UNSPECIFIED;
+    this.gem = opts.gem ?? null;
+
+    if (opts.metadata) this.metadata = opts.metadata;
+    if (opts.cid) this.cid = opts.cid;
+    if (opts.rid) this.rid = opts.rid;
+    if (opts.rcid) this.rcid = opts.rcid;
+  }
+
+  toString(): string {
+    return `ChatSession(cid='${this.cid}', rid='${this.rid}', rcid='${this.rcid}')`;
+  }
+
+  get last_output(): ModelOutput | null {
+    return this._last_output;
+  }
+
+  set last_output(v: ModelOutput | null) {
+    this._last_output = v;
+    if (v) {
+      this.metadata = (v.metadata ?? []) as Array<string | null>;
+      this.rcid = v.rcid;
+    }
+  }
+
+  async send_message(prompt: string, files: string[] | null = null, kwargs: RequestKwargs = {}): Promise<ModelOutput> {
+    return await this.geminiclient.generate_content(prompt, files, this.model, this.gem, this, kwargs);
+  }
+
+  async sendMessage(prompt: string, files?: string[] | null, kwargs?: RequestKwargs): Promise<ModelOutput> {
+    return await this.send_message(prompt, files ?? null, kwargs ?? {});
+  }
+
+  choose_candidate(index: number): ModelOutput {
+    if (!this.last_output) throw new Error('No previous output data found in this chat session.');
+    if (index >= this.last_output.candidates.length) {
+      throw new Error(`Index ${index} exceeds the number of candidates in last model output.`);
+    }
+    this.last_output.chosen = index;
+    this.rcid = this.last_output.rcid;
+    return this.last_output;
+  }
+
+  chooseCandidate(index: number): ModelOutput {
+    return this.choose_candidate(index);
+  }
+
+  get metadata(): Array<string | null> {
+    return this.__metadata;
+  }
+
+  set metadata(v: Array<string | null>) {
+    if (v.length > 3) throw new Error('metadata cannot exceed 3 elements');
+    this.__metadata = [null, null, null];
+    for (let i = 0; i < v.length; i++) this.__metadata[i] = v[i] ?? null;
+  }
+
+  get cid(): string | null {
+    return this.__metadata[0];
+  }
+
+  set cid(v: string | null) {
+    this.__metadata[0] = v;
+  }
+
+  get rid(): string | null {
+    return this.__metadata[1];
+  }
+
+  set rid(v: string | null) {
+    this.__metadata[1] = v;
+  }
+
+  get rcid(): string | null {
+    return this.__metadata[2];
+  }
+
+  set rcid(v: string | null) {
+    this.__metadata[2] = v;
+  }
+}
@@ -0,0 +1,179 @@
+import { GRPC } from '../constants.js';
+import { APIError } from '../exceptions.js';
+import { Gem, GemJar, RPCData } from '../types/index.js';
+import { logger } from '../utils/logger.js';
+import { extract_json_from_response, get_nested_value } from '../utils/parsing.js';
+
+export abstract class GemMixin {
+  protected _gems: GemJar | null = null;
+
+  protected abstract _run<T>(fn: () => Promise<T>, retry: number): Promise<T>;
+  protected abstract _batch_execute(payloads: RPCData[], opts?: RequestInit): Promise<Response>;
+  protected abstract close(delay?: number): Promise<void>;
+
+  get gems(): GemJar {
+    if (this._gems == null) {
+      throw new Error(
+        'Gems not fetched yet. Call `GeminiClient.fetch_gems()` method to fetch gems from gemini.google.com.',
+      );
+    }
+    return this._gems;
+  }
+
+  async fetch_gems(include_hidden: boolean = false, opts?: RequestInit): Promise<GemJar> {
+    return await this._run(async () => {
+      const res = await this._batch_execute(
+        [
+          new RPCData(GRPC.LIST_GEMS, include_hidden ? '[4]' : '[3]', 'system'),
+          new RPCData(GRPC.LIST_GEMS, '[2]', 'custom'),
+        ],
+        opts,
+      );
+
+      let response_json: unknown;
+      try {
+        response_json = extract_json_from_response(await res.text());
+        if (!Array.isArray(response_json)) throw new Error('Invalid response');
+      } catch {
+        await this.close();
+        throw new APIError('Failed to fetch gems. Invalid response data received. Client will try to re-initialize on next request.');
+      }
+
+      let predefined: unknown[] = [];
+      let custom: unknown[] = [];
+
+      try {
+        for (const part of response_json as unknown[]) {
+          if (!Array.isArray(part)) continue;
+          const ident = part[part.length - 1];
+          const body = get_nested_value<string | null>(part, [2], null);
+          if (!body) continue;
+
+          if (ident === 'system') {
+            const parsed = JSON.parse(body) as unknown[];
+            predefined = (Array.isArray(parsed) ? (parsed[2] as unknown[]) : []) ?? [];
+          } else if (ident === 'custom') {
+            const parsed = JSON.parse(body) as unknown[] | null;
+            if (parsed) custom = (parsed[2] as unknown[]) ?? [];
+          }
+        }
+
+        if (predefined.length === 0 && custom.length === 0) throw new Error('No gems');
+      } catch {
+        await this.close();
+        logger.debug('Invalid response while parsing gems');
+        throw new APIError('Failed to fetch gems. Invalid response data received. Client will try to re-initialize on next request.');
+      }
+
+      const entries: [string, Gem][] = [];
+
+      for (const gem of predefined) {
+        if (!Array.isArray(gem)) continue;
+        const id = String(get_nested_value(gem, [0], ''));
+        if (!id) continue;
+        entries.push([
+          id,
+          new Gem(
+            id,
+            String(get_nested_value(gem, [1, 0], '')),
+            get_nested_value<string | null>(gem, [1, 1], null),
+            get_nested_value<string | null>(gem, [2, 0], null),
+            true,
+          ),
+        ]);
+      }
+
+      for (const gem of custom) {
+        if (!Array.isArray(gem)) continue;
+        const id = String(get_nested_value(gem, [0], ''));
+        if (!id) continue;
+        entries.push([
+          id,
+          new Gem(
+            id,
+            String(get_nested_value(gem, [1, 0], '')),
+            get_nested_value<string | null>(gem, [1, 1], null),
+            get_nested_value<string | null>(gem, [2, 0], null),
+            false,
+          ),
+        ]);
+      }
+
+      this._gems = new GemJar(entries);
+      return this._gems;
+    }, 2);
+  }
+
+  async create_gem(name: string, prompt: string, description: string = ''): Promise<Gem> {
+    return await this._run(async () => {
+      const payload = JSON.stringify([
+        [
+          name,
+          description,
+          prompt,
+          null,
+          null,
+          null,
+          null,
+          null,
+          0,
+          null,
+          1,
+          null,
+          null,
+          null,
+          [],
+        ],
+      ]);
+
+      const res = await this._batch_execute([new RPCData(GRPC.CREATE_GEM, payload)]);
+      try {
+        const response_json = extract_json_from_response(await res.text()) as unknown[];
+        const gem_id = JSON.parse(String((response_json[0] as unknown[])[2]))[0] as string;
+        return new Gem(gem_id, name, description, prompt, false);
+      } catch {
+        await this.close();
+        throw new APIError('Failed to create gem. Invalid response data received. Client will try to re-initialize on next request.');
+      }
+    }, 2);
+  }
+
+  async update_gem(gem: Gem | string, name: string, prompt: string, description: string = ''): Promise<Gem> {
+    return await this._run(async () => {
+      const gem_id = typeof gem === 'string' ? gem : gem.id;
+      const payload = JSON.stringify([
+        gem_id,
+        [
+          name,
+          description,
+          prompt,
+          null,
+          null,
+          null,
+          null,
+          null,
+          0,
+          null,
+          1,
+          null,
+          null,
+          null,
+          [],
+          0,
+        ],
+      ]);
+
+      await this._batch_execute([new RPCData(GRPC.UPDATE_GEM, payload)]);
+      return new Gem(gem_id, name, description, prompt, false);
+    }, 2);
+  }
+
+  async delete_gem(gem: Gem | string, opts?: RequestInit): Promise<void> {
+    return await this._run(async () => {
+      const gem_id = typeof gem === 'string' ? gem : gem.id;
+      const payload = JSON.stringify([gem_id]);
+      await this._batch_execute([new RPCData(GRPC.DELETE_GEM, payload)], opts);
+    }, 2);
+  }
+}
+
@@ -0,0 +1,2 @@
+export { GemMixin } from './gem-mixin.js';
+
@@ -0,0 +1,107 @@
+export const Endpoint = {
+  GOOGLE: 'https://www.google.com',
+  INIT: 'https://gemini.google.com/app',
+  GENERATE:
+    'https://gemini.google.com/_/BardChatUi/data/assistant.lamda.BardFrontendService/StreamGenerate',
+  ROTATE_COOKIES: 'https://accounts.google.com/RotateCookies',
+  UPLOAD: 'https://content-push.googleapis.com/upload',
+  BATCH_EXEC: 'https://gemini.google.com/_/BardChatUi/data/batchexecute',
+} as const;
+
+export const GRPC = {
+  LIST_CHATS: 'MaZiqc',
+  READ_CHAT: 'hNvQHb',
+  LIST_GEMS: 'CNgdBe',
+  CREATE_GEM: 'oMH3Zd',
+  UPDATE_GEM: 'kHv0Vd',
+  DELETE_GEM: 'UXcSJb',
+} as const;
+
+export const Headers = {
+  GEMINI: {
+    'Content-Type': 'application/x-www-form-urlencoded;charset=utf-8',
+    Host: 'gemini.google.com',
+    Origin: 'https://gemini.google.com',
+    Referer: 'https://gemini.google.com/',
+    'User-Agent':
+      'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
+    'X-Same-Domain': '1',
+  },
+  ROTATE_COOKIES: {
+    'Content-Type': 'application/json',
+  },
+  UPLOAD: {
+    'Push-ID': 'feeds/mcudyrk2a4khkz',
+  },
+} as const;
+
+export const ErrorCode = {
+  TEMPORARY_ERROR_1013: 1013,
+  USAGE_LIMIT_EXCEEDED: 1037,
+  MODEL_INCONSISTENT: 1050,
+  MODEL_HEADER_INVALID: 1052,
+  IP_TEMPORARILY_BLOCKED: 1060,
+} as const;
+
+export class Model {
+  static readonly UNSPECIFIED = new Model('unspecified', {}, false);
+  static readonly G_3_0_PRO = new Model(
+    'gemini-3.0-pro',
+    { 'x-goog-ext-525001261-jspb': '[1,null,null,null,"9d8ca3786ebdfbea",null,null,0,[4]]' },
+    false,
+  );
+  static readonly G_2_5_PRO = new Model(
+    'gemini-2.5-pro',
+    { 'x-goog-ext-525001261-jspb': '[1,null,null,null,"4af6c7f5da75d65d",null,null,0,[4]]' },
+    false,
+  );
+  static readonly G_2_5_FLASH = new Model(
+    'gemini-2.5-flash',
+    { 'x-goog-ext-525001261-jspb': '[1,null,null,null,"9ec249fc9ad08861",null,null,0,[4]]' },
+    false,
+  );
+
+  constructor(
+    public readonly model_name: string,
+    public readonly model_header: Record<string, string>,
+    public readonly advanced_only: boolean,
+  ) {}
+
+  static from_name(name: string): Model {
+    for (const model of [Model.UNSPECIFIED, Model.G_3_0_PRO, Model.G_2_5_PRO, Model.G_2_5_FLASH]) {
+      if (model.model_name === name) return model;
+    }
+
+    throw new Error(
+      `Unknown model name: ${name}. Available models: ${[Model.UNSPECIFIED, Model.G_3_0_PRO, Model.G_2_5_PRO, Model.G_2_5_FLASH]
+        .map((m) => m.model_name)
+        .join(', ')}`,
+    );
+  }
+
+  static from_dict(model_dict: { model_name?: unknown; model_header?: unknown }): Model {
+    if (!model_dict || typeof model_dict !== 'object') {
+      throw new Error("When passing a custom model as a dictionary, 'model_name' and 'model_header' keys must be provided.");
+    }
+
+    if (!('model_name' in model_dict) || !('model_header' in model_dict)) {
+      throw new Error("When passing a custom model as a dictionary, 'model_name' and 'model_header' keys must be provided.");
+    }
+
+    if (typeof model_dict.model_name !== 'string' || !model_dict.model_name.trim()) {
+      throw new Error("When passing a custom model as a dictionary, 'model_name' must be a non-empty string.");
+    }
+
+    if (!model_dict.model_header || typeof model_dict.model_header !== 'object') {
+      throw new Error("When passing a custom model as a dictionary, 'model_header' must be a dictionary containing valid header strings.");
+    }
+
+    const header: Record<string, string> = {};
+    for (const [k, v] of Object.entries(model_dict.model_header as Record<string, unknown>)) {
+      if (typeof v === 'string') header[k] = v;
+    }
+
+    return new Model(model_dict.model_name, header, false);
+  }
+}
+
@@ -0,0 +1,56 @@
+export class AuthError extends Error {
+  constructor(message = 'AuthError') {
+    super(message);
+    this.name = 'AuthError';
+  }
+}
+
+export class APIError extends Error {
+  constructor(message = 'APIError') {
+    super(message);
+    this.name = 'APIError';
+  }
+}
+
+export class ImageGenerationError extends APIError {
+  constructor(message = 'ImageGenerationError') {
+    super(message);
+    this.name = 'ImageGenerationError';
+  }
+}
+
+export class GeminiError extends Error {
+  constructor(message = 'GeminiError') {
+    super(message);
+    this.name = 'GeminiError';
+  }
+}
+
+export class TimeoutError extends GeminiError {
+  constructor(message = 'TimeoutError') {
+    super(message);
+    this.name = 'TimeoutError';
+  }
+}
+
+export class UsageLimitExceeded extends GeminiError {
+  constructor(message = 'UsageLimitExceeded') {
+    super(message);
+    this.name = 'UsageLimitExceeded';
+  }
+}
+
+export class ModelInvalid extends GeminiError {
+  constructor(message = 'ModelInvalid') {
+    super(message);
+    this.name = 'ModelInvalid';
+  }
+}
+
+export class TemporarilyBlocked extends GeminiError {
+  constructor(message = 'TemporarilyBlocked') {
+    super(message);
+    this.name = 'TemporarilyBlocked';
+  }
+}
+
@@ -0,0 +1,8 @@
+export { GeminiClient, ChatSession } from './client.js';
+
+export * from './exceptions.js';
+export * from './types/index.js';
+export * from './constants.js';
+export { logger, set_log_level, setLogLevel } from './utils/logger.js';
+export * as utils from './utils/index.js';
+
@@ -0,0 +1,45 @@
+import { GeneratedImage, type Image, WebImage } from './image.js';
+
+function decode_html(s: string | null | undefined): string | null | undefined {
+  if (s == null) return s;
+  return s
+    .replace(/&amp;/g, '&')
+    .replace(/&lt;/g, '<')
+    .replace(/&gt;/g, '>')
+    .replace(/&quot;/g, '"')
+    .replace(/&apos;/g, "'")
+    .replace(/&#39;/g, "'")
+    .replace(/&nbsp;/g, ' ')
+    .replace(/&#x([0-9a-fA-F]+);/g, (_, hex) => String.fromCodePoint(parseInt(hex, 16)))
+    .replace(/&#(\d+);/g, (_, dec) => String.fromCodePoint(parseInt(dec, 10)));
+}
+
+export class Candidate {
+  public rcid: string;
+  public text: string;
+  public thoughts: string | null;
+  public web_images: WebImage[];
+  public generated_images: GeneratedImage[];
+
+  constructor(params: {
+    rcid: string;
+    text: string;
+    thoughts?: string | null;
+    web_images?: WebImage[];
+    generated_images?: GeneratedImage[];
+  }) {
+    this.rcid = params.rcid;
+    this.text = decode_html(params.text) ?? '';
+    this.thoughts = decode_html(params.thoughts) ?? null;
+    this.web_images = params.web_images ?? [];
+    this.generated_images = params.generated_images ?? [];
+  }
+
+  toString(): string {
+    return this.text;
+  }
+
+  get images(): Image[] {
+    return [...this.web_images, ...this.generated_images];
+  }
+}
@@ -0,0 +1,75 @@
+export class Gem {
+  constructor(
+    public id: string,
+    public name: string,
+    public description: string | null,
+    public prompt: string | null,
+    public predefined: boolean,
+  ) {}
+
+  toString(): string {
+    return `Gem(id='${this.id}', name='${this.name}', description='${this.description}', prompt='${this.prompt}', predefined=${this.predefined})`;
+  }
+}
+
+export class GemJar implements Iterable<Gem> {
+  private m = new Map<string, Gem>();
+
+  constructor(entries?: Iterable<[string, Gem]>) {
+    if (entries) for (const [id, gem] of entries) this.m.set(id, gem);
+  }
+
+  [Symbol.iterator](): Iterator<Gem> {
+    return this.m.values();
+  }
+
+  entries(): IterableIterator<[string, Gem]> {
+    return this.m.entries();
+  }
+
+  values(): IterableIterator<Gem> {
+    return this.m.values();
+  }
+
+  has(id: string): boolean {
+    return this.m.has(id);
+  }
+
+  set(id: string, gem: Gem): this {
+    this.m.set(id, gem);
+    return this;
+  }
+
+  get(id?: string | null, name?: string | null, def: Gem | null = null): Gem | null {
+    if (id == null && name == null) {
+      throw new Error('At least one of gem id or name must be provided.');
+    }
+
+    if (id != null) {
+      const g = this.m.get(id) ?? null;
+      if (!g) return def;
+      if (name != null) return g.name === name ? g : def;
+      return g;
+    }
+
+    if (name != null) {
+      for (const g of this.m.values()) {
+        if (g.name === name) return g;
+      }
+      return def;
+    }
+
+    return def;
+  }
+
+  filter(predefined: boolean | null = null, name: string | null = null): GemJar {
+    const out: [string, Gem][] = [];
+    for (const [id, gem] of this.m.entries()) {
+      if (predefined != null && gem.predefined !== predefined) continue;
+      if (name != null && gem.name !== name) continue;
+      out.push([id, gem]);
+    }
+    return new GemJar(out);
+  }
+}
+
@@ -0,0 +1,16 @@
+export class RPCData {
+  constructor(
+    public rpcid: string,
+    public payload: string,
+    public identifier: string = 'generic',
+  ) {}
+
+  toString(): string {
+    return `GRPC(rpcid='${this.rpcid}', payload='${this.payload}', identifier='${this.identifier}')`;
+  }
+
+  serialize(): unknown[] {
+    return [this.rpcid, this.payload, null, this.identifier];
+  }
+}
+
@@ -0,0 +1,114 @@
+import path from 'node:path';
+import { mkdir, writeFile } from 'node:fs/promises';
+
+import { logger } from '../utils/logger.js';
+import { cookie_header, fetch_with_timeout } from '../utils/http.js';
+
+export class Image {
+  constructor(
+    public url: string,
+    public title = '[Image]',
+    public alt = '',
+    public proxy: string | null = null,
+  ) {}
+
+  toString(): string {
+    const u = this.url.length <= 20 ? this.url : `${this.url.slice(0, 8)}...${this.url.slice(-12)}`;
+    return `Image(title='${this.title}', alt='${this.alt}', url='${u}')`;
+  }
+
+  async save(
+    p: string = 'temp',
+    filename: string | null = null,
+    cookies: Record<string, string> | null = null,
+    verbose: boolean = false,
+    skip_invalid_filename: boolean = false,
+  ): Promise<string | null> {
+    filename = filename ?? this.url.split('/').pop()?.split('?')[0] ?? 'image';
+    const m = filename.match(/^(.*\.\w+)/);
+    if (m) filename = m[1]!;
+    else {
+      if (verbose) logger.warning(`Invalid filename: ${filename}`);
+      if (skip_invalid_filename) return null;
+    }
+
+    const headers: Record<string, string> = {
+      'User-Agent': 'Mozilla/5.0',
+      Accept: 'image/avif,image/webp,image/apng,image/*,*/*;q=0.8',
+      Referer: 'https://gemini.google.com/',
+    };
+    if (cookies) headers.Cookie = cookie_header(cookies);
+
+    let url = this.url;
+    let res: Response | null = null;
+    for (let i = 0; i < 10; i++) {
+      res = await fetch_with_timeout(url, {
+        method: 'GET',
+        headers,
+        redirect: 'manual',
+        timeout_ms: 30_000,
+      });
+
+      if (res.status >= 300 && res.status < 400) {
+        const loc = res.headers.get('location');
+        if (!loc) break;
+        url = new URL(loc, url).toString();
+        continue;
+      }
+
+      break;
+    }
+
+    if (!res) throw new Error('Image download failed: no response');
+
+    if (!res.ok) {
+      throw new Error(`Error downloading image: ${res.status} ${res.statusText}`);
+    }
+
+    const ct = res.headers.get('content-type');
+    if (ct && !ct.includes('image')) {
+      logger.warning(`Content type of ${filename} is not image, but ${ct}.`);
+    }
+
+    const dir = path.resolve(p);
+    await mkdir(dir, { recursive: true });
+
+    const dest = path.join(dir, filename);
+    const buf = Buffer.from(await res.arrayBuffer());
+    await writeFile(dest, buf);
+
+    if (verbose) logger.info(`Image saved as ${dest}`);
+    return dest;
+  }
+}
+
+export class WebImage extends Image {}
+
+export class GeneratedImage extends Image {
+  constructor(
+    url: string,
+    title: string,
+    alt: string,
+    proxy: string | null,
+    public cookies: Record<string, string>,
+  ) {
+    super(url, title, alt, proxy);
+    if (!cookies || Object.keys(cookies).length === 0) {
+      throw new Error('GeneratedImage is designed to be initialized with same cookies as GeminiClient.');
+    }
+  }
+
+  async save(
+    p: string = 'temp',
+    filename: string | null = null,
+    cookies: Record<string, string> | null = null,
+    verbose: boolean = false,
+    skip_invalid_filename: boolean = false,
+    full_size: boolean = true,
+  ): Promise<string | null> {
+    const u = full_size ? `${this.url}=s2048` : this.url;
+    const f = filename ?? `${new Date().toISOString().replace(/[-:.TZ]/g, '').slice(0, 14)}_${u.slice(-10)}.png`;
+    const img = new Image(u, this.title, this.alt, this.proxy);
+    return await img.save(p, f, cookies ?? this.cookies, verbose, skip_invalid_filename);
+  }
+}
@@ -0,0 +1,6 @@
+export { Candidate } from './candidate.js';
+export { Gem, GemJar } from './gem.js';
+export { RPCData } from './grpc.js';
+export { GeneratedImage, Image, WebImage } from './image.js';
+export { ModelOutput } from './modeloutput.js';
+
@@ -0,0 +1,35 @@
+import type { Image } from './image.js';
+import type { Candidate } from './candidate.js';
+
+export class ModelOutput {
+  public metadata: string[];
+  public candidates: Candidate[];
+  public chosen: number;
+
+  constructor(params: { metadata: string[]; candidates: Candidate[]; chosen?: number }) {
+    this.metadata = params.metadata;
+    this.candidates = params.candidates;
+    this.chosen = params.chosen ?? 0;
+  }
+
+  toString(): string {
+    return this.text;
+  }
+
+  get text(): string {
+    return this.candidates[this.chosen]?.text ?? '';
+  }
+
+  get thoughts(): string | null {
+    return this.candidates[this.chosen]?.thoughts ?? null;
+  }
+
+  get images(): Image[] {
+    return this.candidates[this.chosen]?.images ?? [];
+  }
+
+  get rcid(): string {
+    return this.candidates[this.chosen]?.rcid ?? '';
+  }
+}
+
@@ -0,0 +1,82 @@
+import fs from 'node:fs';
+import path from 'node:path';
+import { mkdir, readFile, writeFile } from 'node:fs/promises';
+
+import { resolveGeminiWebCookiePath } from './paths.js';
+
+export type CookieMap = Record<string, string>;
+
+export type CookieFileData =
+  | {
+      cookies: CookieMap;
+      updated_at: number;
+      source?: string;
+    }
+  | {
+      version: number;
+      updatedAt: string;
+      cookieMap: CookieMap;
+      source?: string;
+    };
+
+export async function read_cookie_file(p: string = resolveGeminiWebCookiePath()): Promise<CookieMap | null> {
+  try {
+    if (!fs.existsSync(p) || !fs.statSync(p).isFile()) return null;
+    const raw = await readFile(p, 'utf8');
+    const data = JSON.parse(raw) as unknown;
+
+    if (data && typeof data === 'object' && 'cookies' in (data as any)) {
+      const cookies = (data as any).cookies as unknown;
+      if (cookies && typeof cookies === 'object') {
+        const out: CookieMap = {};
+        for (const [k, v] of Object.entries(cookies as Record<string, unknown>)) {
+          if (typeof v === 'string') out[k] = v;
+        }
+        return out;
+      }
+    }
+
+    if (data && typeof data === 'object' && 'cookieMap' in (data as any)) {
+      const cookies = (data as any).cookieMap as unknown;
+      if (cookies && typeof cookies === 'object') {
+        const out: CookieMap = {};
+        for (const [k, v] of Object.entries(cookies as Record<string, unknown>)) {
+          if (typeof v === 'string') out[k] = v;
+        }
+        return Object.keys(out).length > 0 ? out : null;
+      }
+    }
+
+    if (data && typeof data === 'object') {
+      const out: CookieMap = {};
+      for (const [k, v] of Object.entries(data as Record<string, unknown>)) {
+        if (typeof v === 'string') out[k] = v;
+      }
+      return Object.keys(out).length > 0 ? out : null;
+    }
+
+    return null;
+  } catch {
+    return null;
+  }
+}
+
+export async function write_cookie_file(
+  cookies: CookieMap,
+  p: string = resolveGeminiWebCookiePath(),
+  source?: string,
+): Promise<void> {
+  const dir = path.dirname(p);
+  await mkdir(dir, { recursive: true });
+
+  const payload: CookieFileData = {
+    version: 1,
+    updatedAt: new Date().toISOString(),
+    cookieMap: cookies,
+    source,
+  };
+  await writeFile(p, JSON.stringify(payload, null, 2), 'utf8');
+}
+
+export const readCookieFile = read_cookie_file;
+export const writeCookieFile = write_cookie_file;
@@ -0,0 +1,34 @@
+import { APIError, ImageGenerationError } from '../exceptions.js';
+import { sleep } from './http.js';
+
+export function running(retry: number = 0) {
+  return <TArgs extends unknown[], TResult>(
+    fn: (client: any, ...args: TArgs) => Promise<TResult>,
+  ): ((client: any, ...args: TArgs) => Promise<TResult>) => {
+    const wrap = async (client: any, ...args: TArgs): Promise<TResult> => {
+      try {
+        if (!client?._running) {
+          await client.init?.({
+            timeout: client.timeout,
+            auto_close: client.auto_close,
+            close_delay: client.close_delay,
+            auto_refresh: client.auto_refresh,
+            refresh_interval: client.refresh_interval,
+            verbose: false,
+          });
+        }
+        return await fn(client, ...args);
+      } catch (e) {
+        let r = retry;
+        if (e instanceof ImageGenerationError) r = Math.min(1, r);
+        if (e instanceof APIError && r > 0) {
+          await sleep(1000);
+          return await running(r - 1)(fn)(client, ...args);
+        }
+        throw e;
+      }
+    };
+    return wrap;
+  };
+}
+
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Jim Liu 宝玉	7a193481d4	chore: release v1.4.0	2026-01-18 23:46:07 -06:00
Jim Liu 宝玉	aea27950ad	chore: release v1.3.0	2026-01-18 22:16:24 -06:00
Jim Liu 宝玉	76ddcf134e	chore: release v1.2.0	2026-01-18 20:42:40 -06:00
Jim Liu 宝玉	5a0feba96e	docs: update installation instructions and add available plugins section	2026-01-18 20:28:36 -06:00
Jim Liu 宝玉	97bc68efd8	chore: release v1.1.0	2026-01-18 20:11:49 -06:00
Jim Liu 宝玉	464edf0656	chore: release v1.0.1	2026-01-18 19:15:34 -06:00
Jim Liu 宝玉	43c2589a38	Refactor code structure for improved readability and maintainability	2026-01-18 18:18:54 -06:00
Jim Liu 宝玉	4c82884722	chore: release v1.0.0 - Add baoyu-danger-x-to-markdown skill to marketplace - Update README files with new skill documentation - Bump version to 1.0.0 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>	2026-01-18 18:00:58 -06:00
Jim Liu 宝玉	56d0485412	chore: rename baoyu-gemini-web to baoyu-danger-gemini-web	2026-01-18 17:58:12 -06:00
Jim Liu 宝玉	4998eaf8c2	feat: add baoyu-danger-x-to-markdown skill Ignore generated x-to-markdown output directory.	2026-01-18 17:42:25 -06:00
Jim Liu 宝玉	5993b6969d	chore: release v0.11.0	2026-01-18 09:53:48 -06:00
Jim Liu 宝玉	dc0201b63f	chore: release v0.10.0	2026-01-18 00:01:06 -06:00
Jim Liu 宝玉	3811512750	docs: add Customization section to README files	2026-01-17 22:12:01 -06:00
Jim Liu 宝玉	9a9f6a42cd	chore: release v0.9.0	2026-01-17 22:02:40 -06:00
Jim Liu 宝玉	3b13b25f1a	chore: release v0.8.2	2026-01-17 19:00:27 -06:00
Jim Liu 宝玉	688d1760ed	chore: release v0.8.0 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>	2026-01-17 16:44:50 -06:00
Jim Liu 宝玉	a701be873b	update Image Generation Guidelines	2026-01-17 16:24:23 -06:00
Jim Liu 宝玉	1df5e4974d	feat: add analysis framework for Xiaohongshu content	2026-01-17 15:52:00 -06:00
Jim Liu 宝玉	2677e730b9	Update release-skills to update readme	2026-01-17 15:11:50 -06:00
Jim Liu 宝玉	080f2eff48	chore: release v0.7.0 - baoyu-comic: adds --aspect (3:4, 4:3, 16:9) and --lang options; multi-variant storyboard workflow - baoyu-comic/baoyu-slide-deck: adds analysis-framework and template references - Multiple skills: restructured SKILL.md, moved details to references/ Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>	2026-01-17 15:03:43 -06:00
Jim Liu 宝玉	bb4f0dc52c	feat: add PDF export for slide-deck and comic skills	2026-01-17 12:08:08 -06:00
Jim Liu 宝玉	c731faea8f	feat: update slide deck	2026-01-17 04:28:16 -06:00
Jim Liu 宝玉	fa155da15d	fix: update version to 0.5.3 and enhance placeholder selection logic in publishArticle function	2026-01-17 01:29:10 -06:00
				`@@ -0,0 +1,2 @@`
				`export { GemMixin } from './gem-mixin.js';`