← 返回 Skills 市场
hzhyjr2021-beep

city-life-copilot

作者 qiuhe · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
72
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install city-life-copilot
功能描述
告别冰冷导航的全场景城市管家。结合高德地图与网页抓取,提供情绪盲盒、一键拔草、房产雷达与无障碍守护服务。
使用说明 (SKILL.md)

City Life Copilot (城市生活智能管家)

这不是一个普通的地图工具,而是一个能听懂你的情绪、帮你一键转化种草笔记、甚至能自动排雷无障碍路线的超级城市大脑

基于高德地图 + 智能网页抓取,为你生成带有动态地图的高颜值本地 HTML 卡片。


四大杀手锏功能

🎁 情绪盲盒漫游

告诉它你累了、心情不好、想一个人静静,它自动为你生成A/B 双线治愈路线

  • 方案 A · 艺术治愈之旅(室内美术馆 + 书店)
  • 方案 B · 自然呼吸之旅(公园漫步 + 湖边散步)

无需查天气、无需做攻略,把决策交给它,你只管出发。

🔗 种草一键拔草

丢给它一篇小红书/社交媒体种草链接,自动:

  1. 使用 agent-browser 抓取笔记内容(⚠️ 小红书有强反爬虫机制,web-fetch 会返回 error_code=300012)
  2. 提取笔记中的所有地点
  3. 调用高德 API 清洗坐标
  4. 分段计算步行/驾车路线
  5. 生成连贯的动态地图路线

拒绝绕路费腿,博主同款路线一键复刻。

⚠️ 重要说明:

  • 小红书、抖音等强反爬虫平台必须使用 agent-browser
  • web-fetch 会被拦截并返回 error_code=300012 (IP at risk)
  • 脚本 scripts/xiaohongshu-grabber.js 专门处理此类链接

🏠 硬核房产雷达

输入租房/买房意图(如"看中了 XX 小区,在 YY 地上班"),自动触发五维雷达扫描

  • 🚇 交通便利度(地铁站/公交枢纽)
  • 🏥 医疗保障(三甲医院/社区卫生服务中心)
  • 🎓 教育资源(幼儿园/小学)
  • 🛍️ 商业休闲(大型商场/咖啡厅/电影院)
  • 🌳 生态环境(公园/绿地)

附带早高峰真实通勤测算(地铁/公交/打车多方案对比),生成 HTML 体检报告卡片。

♿ 无障碍守护

专为轮椅、婴儿车、银发族设计的避坑 + 绿洲路线

  • 默认体力边界:连续推行不超过 15 分钟(约 800 米)
  • 默认接驳方式:中短途平地推行,长途强制建议打车
  • 自动抓取网友避雷攻略(哪个门有台阶、哪个通道坡太陡)
  • 沿途标注绿洲补给站(带卫生间的商场/咖啡厅)

这不是最快的路线,而是最温柔的路线。


优雅交付

所有结果均输出为高颜值、浅色系本地 HTML 卡片

  • 内嵌动态地图(高德 travel_plan.html + URI 编码)
  • 现代化 Tailwind CSS 排版
  • 无 Emoji 干扰,极致克制的高级感
  • 文件保存在本地 workspace,可离线查看

安装指南

只需一键安装,系统会自动配置所需的高德与网页抓取依赖:

# 方式一:使用技能安装脚本(推荐)
cd /path/to/city-life-copilot
node scripts/install.js

# 方式二:手动安装核心依赖
npx skills add https://www.modelscope.cn/skills/@AMap-Web/amap-lbs-skill
skillhub install agent-browser
# web-fetch 会在运行 install.js 时自动从 ClawHub 安装

核心依赖(已精简至 3 个):

  • amap-lbs-skill — 高德地图综合服务(POI 搜索/路径规划)
  • agent-browser — 浏览器自动化(复杂页面抓取)
  • web-fetch — 网页内容抓取(反爬虫,从 ClawHub 自动安装)

文件结构

city-life-copilot/
├── SKILL.md              # 技能主定义文件
├── scripts/
│   ├── install.js        # 依赖检查与安装脚本
│   ├── dispatcher.js     # 主调度引擎(四重人格路由)
│   └── xiaohongshu-grabber.js  # 小红书链接抓取器(绕过反爬虫)
├── references/
│   └── workflows.md      # 四大场景工作流规则
└── assets/
    └── templates/
        ├── tpl_blind_box.html    # 情绪盲盒专属模板
        ├── tpl_house_radar.html  # 房产雷达专属模板
        └── tpl_general_route.html # 通用及无障碍路线模板

使用示例

# 情绪盲盒
"最近加班好累,周末想一个人出去走走"

# 一键拔草
"帮我规划这个路线:https://www.xiaohongshu.com/..."

# 房产评估
"我打算租大山子南里的房子,在学院国际大厦上班,帮我评估一下"

# 无障碍出行
"我推着婴儿车,从学院国际大厦到元大都遗址公园看海棠花,怎么走最方便"

版本

v1.0.0 — 初始版本,四重人格核心功能上线


让城市生活更简单,更有温度。

安全使用建议
This skill implements the described city assistant, but several red flags mean you should be cautious before installing or running it: - Do NOT run scripts/install.js or any included installer without review. It runs npm -g, git clone, curl/unzip and writes into your home directory and global npm space. - The skill uses AMap (高德) APIs but does not declare where to supply the required API key. Confirm how/where you must provide AMap credentials and never paste secrets into untrusted scripts. - The installer downloads a 'web-fetch' tool from a non-standard host (ClawHub). Verify that URL and the upstream project source before allowing downloads; prefer official release pages (GitHub releases, vendor sites) or inspect archives in a sandbox. - The skill explicitly suggests bypassing anti‑scraping protections with agent-browser. Scraping protected content can violate site terms of service and may carry legal/ethical risk. - If you want to try it, first run it in an isolated environment (VM or container) and inspect network activity. Alternatively, manually install only the dependencies you trust (amap skill from a known source) and avoid automatically running install.js. What would change this assessment: an explicit install manifest that uses only vetted package sources (official releases), clear documentation and prompts for supplying AMap API keys (and no implicit credential access), or provenance for the ClawHub resource (e.g., a reputable GitHub repo or signed release). If the publisher supplies an audited release and documents required credentials and permissions, the rating could move toward benign.
功能分析
Type: OpenClaw Skill Name: city-life-copilot Version: 1.0.0 The skill bundle is classified as suspicious primarily due to the high-risk behavior in `scripts/install.js`, which performs global system modifications and fetches remote code. Specifically, it executes `npm install -g`, and uses `git clone` and `curl` to download and install artifacts from external domains such as `clawhub.ai` (e.g., `https://clawhub.ai/dlutwuwei/web-anti-crawl-fetch`) and `modelscope.cn`. While these actions are presented as necessary dependency installations for the agent's scraping and mapping functions, they represent a significant supply-chain risk and remote code execution (RCE) vector. No explicit evidence of data exfiltration or intentional backdoors was found in the logic of `scripts/dispatcher.js` or the instructions in `SKILL.md`, but the installation pattern exceeds standard safety boundaries for skill bundles.
能力标签
crypto
能力评估
Purpose & Capability
Name/description align with the shipped files: dispatcher, grabber, templates all implement route planning, A/B mood routes, house radar and accessible routing. However, the skill clearly expects to call high‑privilege external services (AMap APIs) but does not declare any required API keys/credentials in its metadata; that omission is inconsistent with the claimed capabilities because geocoding/route APIs normally require a key.
Instruction Scope
SKILL.md and scripts instruct the agent to perform web scraping (including bypassing strong anti‑scraping measures) and to write HTML outputs to a fixed user workspace (~/.openclaw/workspace). The code insists on using an 'agent-browser' to bypass anti‑crawl protections for small‑red‑book (小红书) and similar sites. The workflow has automatic scanning behaviors (e.g., 'hardcore steward' mode triggers active 2km scans without explicit user confirmation). The instructions also tell the agent to install and run local tools and to save files under the user's home directory — these actions go beyond a read‑only assistant and should be made explicit to the user.
Install Mechanism
There is no platform install spec but an included scripts/install.js is a full installer that executes system commands: npm install -g, 'skillhub' commands, 'npx skills add', git clone and curl/unzip fallbacks, and copies files into ~/.openclaw. It fetches code from third‑party endpoints (ModelScope is plausible, but ClawHub at https://clawhub.ai/dlutwuwei/web-anti-crawl-fetch is not a standard vetted release host). The installer performs network downloads and writes into the user's filesystem and may install global npm packages — this is higher risk and should be reviewed before running.
Credentials
The skill declares no required environment variables, but its functionality depends on external services (AMap) that typically require API keys. The manifest does not request or document where to provide AMap credentials or other secrets; at runtime those credentials will still be necessary (or the skill will fail). Additionally, the installer pulls third‑party scraping tools and agent-browser which may in turn require credentials or elevated privileges; the absence of explicit credential requirements is disproportionate and understates what the user must provide to operate the skill.
Persistence & Privilege
always:false (no forced global enable), but the installer will modify user state: create ~/.openclaw workspaces/skills, install SkillHub globally (npm -g), and clone/copy third‑party skill code into the workspace. The skill does not set always:true or modify other skills' configs directly in the repo, but running its install script grants it the ability to change system/global state — user consent and review are needed before running.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install city-life-copilot
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /city-life-copilot 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
City Life Copilot v1.0.0 — Initial release with four core urban assistant features: - Mood Box Roaming: Generates soothing city routes based on your mood via A/B plans (art or nature). - Instant Route Replication: Converts social media “must-visit” posts into real local routes, with mapping and anti-crawling strategies. - Housing Radar: Assesses properties with five-dimension scans (transport, medical, education, lifestyle, environment) and peak-hour commute analysis. - Accessibility Guardian: Plans gentle, accessible routes for strollers, wheelchairs, and the elderly, including pitfall avoidance and rest stops. - All results output as stylish, local HTML cards with embedded maps; streamlined installation process included.
元数据
Slug city-life-copilot
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

city-life-copilot 是什么?

告别冰冷导航的全场景城市管家。结合高德地图与网页抓取,提供情绪盲盒、一键拔草、房产雷达与无障碍守护服务。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 72 次。

如何安装 city-life-copilot?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install city-life-copilot」即可一键安装,无需额外配置。

city-life-copilot 是免费的吗?

是的,city-life-copilot 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

city-life-copilot 支持哪些平台?

city-life-copilot 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 city-life-copilot?

由 qiuhe(@hzhyjr2021-beep)开发并维护,当前版本 v1.0.0。

💬 留言讨论