本文へ移動
cccskills
無料GitHub で公開

glm-vision

This skill should be used when the user sends an image and asks to "analyze this image", "describe this picture", "what's in this image", or any request requiring visual understanding of images. Provides image analysis using Zhipu GLM-4.6V multimodal model.

インストール方法を見る

含まれるファイル(3)

  • SKILL.md4.1 KB
  • references/api-reference.md4.5 KB
  • scripts/analyze_image.py4.4 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

GLM Vision - 图片分析技能

概述

使用智谱 GLM-4.6V 多模态模型分析图片内容。支持:

  • 图片内容描述和理解
  • 图片中的文字提取(OCR)
  • 图片元素识别和分析
  • 多图对比分析
  • 视频内容理解

使用场景

当用户发送图片并提出以下类型问题时触发:

  • "这张图片里有什么?"
  • "请描述这张图片"
  • "分析这个截图"
  • "这是什么?"
  • "图片里的文字是什么?"
  • "比较这两张图片"

API 配置

环境变量

需要设置 ZHIPU_API_KEY 环境变量:

export ZHIPU_API_KEY="your-api-key-here"

或者在 ~/.claude/settings.json 中配置:

{
  "env": {
    "ZHIPU_API_KEY": "your-api-key-here"
  }
}

模型选择

模型用途价格
glm-4.6v-flash免费版,日常使用免费
glm-4.6v旗舰版,复杂推理付费
glm-4.6v-flashx轻量高速版付费

默认使用 glm-4.6v-flash 免费模型。

调用方式

方式一:使用辅助脚本(推荐)

python3 ~/.claude/skills/glm-vision/scripts/analyze_image.py \
  --image "/path/to/image.jpg" \
  --prompt "请描述这张图片" \
  [--model "glm-4.6v-flash"]

参数说明:

  • --image / -i: 图片路径(支持 URL 或本地文件)
  • --prompt / -p: 分析提示词
  • --model / -m: 模型名称(可选,默认免费版)
  • --detail / -d: 启用详细思考模式

方式二:直接 API 调用

import base64
from openai import OpenAI

client = OpenAI(
    api_key=os.environ.get("ZHIPU_API_KEY"),
    base_url="https://open.bigmodel.cn/api/paas/v4"
)

# 读取本地图片
with open("image.jpg", "rb") as f:
    img_base64 = base64.b64encode(f.read()).decode("utf-8")

response = client.chat.completions.create(
    model="glm-4.6v-flash",
    messages=[{
        "role": "user",
        "content": [
            {"type": "image_url", "image_url": {"url": img_base64}},
            {"type": "text", "text": "请描述这张图片"}
        ]
    }]
)
print(response.choices[0].message.content)

方式三:URL 图片

response = client.chat.completions.create(
    model="glm-4.6v-flash",
    messages=[{
        "role": "user",
        "content": [
            {"type": "image_url", "image_url": {"url": "https://example.com/image.jpg"}},
            {"type": "text", "text": "分析这张图片"}
        ]
    }]
)

工作流程

  1. 接收图片: 用户发送图片,保存到临时目录
  2. 构建请求: 将图片转为 base64 或使用 URL
  3. 调用 API: 发送到 GLM-4.6V 模型
  4. 返回结果: 解析并展示分析结果

支持的图片格式

  • JPEG / JPG
  • PNG
  • GIF(静态)
  • WebP
  • BMP

示例用法

基础图片描述

python3 ~/.claude/skills/glm-vision/scripts/analyze_image.py \
  -i "/tmp/screenshot.png" \
  -p "描述这张截图的内容"

详细分析

python3 ~/.claude/skills/glm-vision/scripts/analyze_image.py \
  -i "/tmp/photo.jpg" \
  -p "分析这张图片的构图、色彩和主题" \
  --detail

OCR 文字提取

python3 ~/.claude/skills/glm-vision/scripts/analyze_image.py \
  -i "/tmp/document.png" \
  -p "提取图片中的所有文字,保持原有格式"

URL 图片分析

python3 ~/.claude/skills/glm-vision/scripts/analyze_image.py \
  -i "https://example.com/image.jpg" \
  -p "这张图片展示的是什么?"

注意事项

  1. API Key 安全: 不要在代码中硬编码 API Key
  2. 图片大小: 建议图片小于 10MB,过大的图片会增加延迟
  3. 网络连接: 需要稳定的网络访问智谱 API
  4. 并发限制: 免费用户并发数为 5

参考资源

  • references/api-reference.md - 完整 API 参考文档
  • scripts/analyze_image.py - 图片分析辅助脚本

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Browser automation CLI for AI agents. Use when needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. TRIGGER when user requests to "open a website", "fill out a form", "click a button", "take a screenshot", "debug this in browser", "scrape data from a page", "test this web app", "login to a site", "frontend UI/UX aesthetics", "automate browser actions", or any task requiring programmatic web interaction.

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

Review common AI slops of defensive programming patterns, avoid silent errors. TRIGGER when reviewing code for defensive anti-patterns, writing fail-fast code, or auditing error handling quality.

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

ast-grep

無料

Guide for writing ast-grep rules to perform structural code search and analysis. This skill should be used when users need to search codebases using Abstract Syntax Tree (AST) patterns, find specific code structures, or perform complex code queries that go beyond simple text search, or when a simple grep/glob search is insufficient for structural code pattern matching.

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

Best practices for AI-driven English-to-Chinese translation. This skill should be used when the user asks to "translate to Chinese", "update the Chinese translation", "improve Chinese translation", "fix translation quality", "review Chinese translation", or when translating any English text into Chinese. Also applies when polishing an existing Chinese translation of English content.

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

This skill should be used when the user asks to "use bilibili API", "download bilibili video", "get bilibili user info", "list bilibili favorites", "send bilibili danmaku", "upload video to bilibili", "monitor bilibili live room", "search bilibili", "get bilibili comments", or needs guidance on the bilibili_api Python library usage, authentication, API endpoints, or workflow patterns.

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

This skill should be used when sending images, files, or notifications back to the user via messaging platforms (Discord, Feishu, Telegram, etc.) through cc-connect. TRIGGER when agent generates a plot/chart/screenshot and wants to show the user; agent creates a report/PDF/file the user should receive; agent needs to proactively notify the user (e.g. task completed, alert, reminder); user asks to "send image", "show me the chart", "notify me", "send the file", "send to Telegram", "show plot in Discord".

日本語の概要は準備中です。原文の説明を表示しています。

archibate/dotfiles-opencode1082026年4月29日 更新

archibate のスキルをすべて見る

このスキルの問題を報告する