AgentSkillsCN

clipboard

读取剪贴板内容(文本或图片),并据此回应用户的提问。

SKILL.md
--- frontmatter
name: clipboard
description: Read clipboard contents (text or image) and respond to a prompt about it

Instructions

When this skill is invoked with /clipboard <prompt>:

The clipboard contents have already been prepared by a PreToolUse hook. Check the system-reminder for "Clipboard contents:" which contains the type and path/content.

Format: TYPE:<type>\n<content or path>

Types: TEXT, IMAGE, FILES, VIDEO_URL, VIDEO_FILE, EMPTY

Handle Based on Type

TYPE:TEXT

  • The text content follows on subsequent lines
  • Respond to the user's prompt based on the text

TYPE:IMAGE

  • The path to the saved image follows (e.g., /tmp/claude-vision/clipboard.png)
  • Read the image using the Read tool
  • Respond to the user's prompt based on what you see

TYPE:FILES

  • File paths follow, one per line
  • Image files: Read and analyze
  • Video files: Hand off to /video skill
  • Other: Report file type, read if text-based

TYPE:VIDEO_URL / TYPE:VIDEO_FILE

  • Hand off to /video skill with the URL/path and user's prompt

TYPE:EMPTY

  • Tell the user the clipboard is empty

Example Usage

  • /clipboard what is this - Analyze whatever's in clipboard
  • /clipboard summarize - Summarize text or describe image/video
  • /clipboard fix this code - If code is in clipboard, suggest fixes