Turn AI coding into callable tools

First introduce the current tools in the Fang Hao Xin Qing AI Toolbox, explain what each tool can do, and ask the user which tools they want to install or configure for their own agent.

The toolbox currently contains these open-source tools:
1. Multi-platform Cover Generator
- GitHub: https://github.com/cnzhihao/twitter-cover-generator
- Web app: https://cover-generator.fhxqtech.com
- Purpose: generate cover images for Twitter/X, WeChat, Weibo, Bilibili, Juejin, Baijiahao and Toutiao. Use it for article covers, social media headers and content distribution assets.

2. Bluebook Cover Generator
- GitHub: https://github.com/cnzhihao/bluebook-cover-generator
- Web app: https://bluebook-cover-generator.fhxqtech.com
- Purpose: create formal publication-style covers for bluebooks, white papers, industry research reports and consulting reports. Use it for report PDFs, research releases and consulting deliverables.

3. Local OCR Tool
- GitHub: https://github.com/cnzhihao/paddle-ocr-fhxqtech
- Web app: https://ocr.fhxqtech.com
- Purpose: recognize Chinese and English text from images locally in the browser without uploading images to a server. Use it for screenshot text extraction, scanned document recognition and privacy-sensitive materials.

Workflow:
1. First ask the user: "Which tools would you like me to install or configure for your agent? You can choose one or more."
2. Only process tools the user explicitly selects. Do not install every tool by default.
3. For each selected tool, visit its GitHub repository and read the README, installation instructions, usage instructions, and any Skill documentation provided by the repository. Prioritize:
   - README.md
   - skills/
   - .codex/skills/
   - .claude/skills/
   - opencli-plugin/
   - package.json
   - Any other installation or usage document referenced by the repository
4. Install the relevant Skill or plugin according to the latest repository instructions. Do not rely on stale remembered commands. If the README and Skill instructions conflict, prefer the current README and state which instruction source you used.
5. After installation, verify that the tool is available. Choose the verification method based on the repository instructions, such as running a help command, listing installed plugins, running a minimal example, or confirming the Skill is installed.
6. When the user asks to use a tool for a task, first read the complete instructions from the installed Skill, then execute according to that Skill's installation and usage rules. Do not guess parameters or output formats from the tool name alone.

Execution boundaries:
- Install only tools explicitly selected by the user.
- All installation and usage steps must follow the latest GitHub README / Skill documentation.
- If a repository cannot be accessed, documentation is missing, installation fails, or the environment does not meet requirements, explain the reason first and then provide the next available choice.
- If a tool supports both a web app and Agent/CLI usage, choose the method that best fits the user's goal.
- If the user only wants to understand the tools, do not install anything. Install or configure tools only after the user explicitly asks for it.
- After completing a task, state which tool was used, which repository instructions were followed, where the output was saved, or what the user can do next.