← 返回 Skills 市场
Planit
作者
yoborlon-alpha
· GitHub ↗
· v2.0.8
· MIT-0
568
总下载
1
收藏
2
当前安装
14
版本数
在 OpenClaw 中安装
/install planit
功能描述
一句话规划出行——输入自然语言即可获得AI优化过的出行方案。触发词：出行、旅游、旅行、行程、规划、路线、度假、游玩、机票、火车票、酒店、景点、攻略
安全使用建议
This skill sends your full message text and userId to a backend server (default: http://8.216.37.65:3721) and will use a PLANIT_SECRET if present, but the package metadata does not disclose these requirements. Before installing, consider: (1) avoid sending any sensitive or personally identifiable information through this skill; (2) only install if you trust the backend server—prefer a named domain and a reviewed endpoint over an IP address; (3) inspect or create the local config file (~/.openclaw/data/planit/config.json) to point to a known endpoint and set PLANIT_SECRET intentionally; (4) ask the author for clarification about telemetry and exactly what is logged/sent (the SECURITY.md claims differ from the code); (5) run the included tests in an isolated environment to observe actual network traffic. If you cannot verify the backend or the author, do not enable this skill for private conversations.
功能分析
Type: OpenClaw Skill
Name: planit
Version: 2.0.8
The skill transmits the full 'skillConfig' object, which often contains sensitive API keys or credentials in the OpenClaw ecosystem, to a hardcoded IP address (8.216.37.65) via both the functional '/plan' endpoint and a '/telemetry' endpoint (src/server-client.js). While the SECURITY.md claims telemetry is anonymous, the code in src/index.js explicitly includes the configuration data in telemetry events, which is a high-risk pattern for potential credential harvesting. The use of a hardcoded IP instead of a domain name further reduces transparency.
能力评估
Purpose & Capability
The skill description promises AI-generated itineraries. Implementing that via a backend API is reasonable, but the package does not declare any required environment variables or config paths while the code reads ~/.openclaw/data/planit/config.json and process.env.PLANIT_SECRET / PLANIT_SERVER_URL. Declaring nothing in metadata but requiring those values at runtime is an incoherence: either the skill should declare and justify those configuration items or it should not rely on them.
Instruction Scope
SKILL.md and SECURITY.md suggest limited, privacy-preserving telemetry and no arbitrary external fetching, but the runtime code actually posts the entire incoming message (including message.text and userId) to a backend /plan endpoint and sends telemetry to /telemetry. The SECURITY.md text contradicts itself (saying 'zero direct external API calls' and then describing POSTs to a backend). The agent will therefore transmit user input and userId to an external server by default—behavior not clearly described or declared in the skill metadata.
Install Mechanism
There is no install spec (lowest install-risk class), but the skill bundle includes runnable Node.js code that will be executed by the agent. The code makes network requests to a default IP (http://8.216.37.65:3721) if no config/env is provided; this is not a remote download at install time, but it is runtime network activity that may have privacy implications.
Credentials
The registry lists no required env vars or config paths, yet the code uses process.env.PLANIT_SECRET and process.env.PLANIT_SERVER_URL and loads ~/.openclaw/data/planit/config.json. The code will include an Authorization Bearer header if PLANIT_SECRET is set and will send user messages and IDs to the remote server. Requiring an undeclared secret and sending potentially sensitive conversation content is disproportionate to the minimal description and is a privacy risk.
Persistence & Privilege
The skill is not marked always:true and does not modify other skills or system-wide settings. It only reads a per-skill config file in the user's home directory and makes network calls; it does not request persistent platform-level privileges in metadata.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install planit - 安装完成后，直接呼叫该 Skill 的名称或使用
/planit触发 - 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
v2.0.8
- Updated skill version to 2.0.8.
- SKILL.md: Updated documentation to reflect the new version number.
v2.0.7
PlanIt 2.0.7 — English Documentation Update
- Updated SKILL.md to provide documentation in English.
- Clarified feature list and usage instructions in concise English examples and descriptions.
- No changes to code or functionality.
v2.0.6
- Added support for configuring a custom server address via config file or environment variables.
- Default server address is now built-in; users can use the skill out of the box without manual configuration.
- Documentation updated with detailed setup instructions for server configuration and priority.
- Minor internal updates and test adjustments for improved reliability.
v2.0.5
No visible changes in this version (2.0.5); SKILL.md and code remain unchanged from previous release.
v2.0.3
**planit v2.0.3 Changelog**
- Updated documentation in SKILL.md: simplified usage, features, and architecture descriptions for improved clarity.
- Revised security guidelines in SECURITY.md.
- Refactored config loading logic in src/config.js for better robustness.
- Minor dependency version bump in package.json.
v2.0.0
PlanIt 2.0.0 — 架构重构，本地仅留薄客户端，核心全面迁移至后端 AI 服务
- 完全重构为“后端 AI Agent + Thin Skill Client”架构
- Skill 不再包含数据抓取与编排，仅负责转发请求和遥测
- 需本地启动 planit-server 后端 (含意图解析、行程生成、数据获取)
- 新增后端遥测与贡献系统
- 个性化酒店排序及多轮对话能力移至服务端
- 运行/测试需手动配置与对接后端 API
v2.0.2
### v2.0.2
- 更新文档版本号至 v2.0.2
- 无功能及架构变更，仅文档和描述更新
v2.0.1
**Major architecture upgrade to Backend AI Agent + Thin Skill Client.**
- Refactored to a thin client: all itinerary logic, intent parsing, and data (12306, 高德) processing now handled by an external backend (planit-server).
- Removed all local data fetching code; Skill only calls backend API and submits telemetry.
- New backend API endpoints: `/plan`, `/telemetry`, `/contributions`; authentication via PLANIT_SECRET.
- Tests now require a running backend server.
- Added configuration for connecting to external server via environment variable or config file.
- Security: Skill can no longer directly access external data sources.
v1.4.0
# planit v1.4.0 Changelog
- Updated dependencies in `package.json`.
- No user-facing functional changes.
v1.3.1
- Bump version to 1.3.1.
- Documentation updates in SKILL.md (version number updated to v1.3.1).
- No functional or API changes.
v1.3.0
**v1.3.0 introduces proxy-based real data fetching for hotels, attractions, and transport.**
- 新增代理服务器支持，自动获取真实交通班次、酒店与景点，不需配置 API key
- 交通、酒店、景点等数据优先走代理服务，无法联网时降级至本地模拟数据
- 文件结构调整：数据抓取逻辑重构为 src/fetcher/ 目录，移除旧 amap.js
- 代理服务地址可通过 config.json 或环境变量灵活配置
- SKILL.md 更新，详细说明数据抓取逻辑与用户零配置体验
v1.2.0
多轮对话引导（城市/天数补问）、实时城市建议、城市库扩展至40+城市
v1.1.0
Automatic 12306 cookie refresh, pure Node.js server (no Chrome/Puppeteer), real train data via queryG endpoint
v1.0.0
PlanIt v1.0.0 — 一句话规划出行，离线版：
- 支持用自然语言一句话获取完整旅行行程，包括交通、酒店推荐（个性化排序）、每日景点时间轴与费用预估
- 意图解析涵盖目的地、时间、人群、预算、时长等，自动生成定制行程
- 内置模拟数据，纯本地处理，无需联网
- 用户每次预订酒店自动记忆，下次同城优先排序已住酒店
- 无依赖、全 Node.js 标准库，跨平台运行
- 自带 28 个用例测试，覆盖常见需求
元数据
常见问题
Planit 是什么？
一句话规划出行——输入自然语言即可获得AI优化过的出行方案。触发词：出行、旅游、旅行、行程、规划、路线、度假、游玩、机票、火车票、酒店、景点、攻略. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 568 次。
如何安装 Planit？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install planit」即可一键安装，无需额外配置。
Planit 是免费的吗？
是的，Planit 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Planit 支持哪些平台？
Planit 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 Planit？
由 yoborlon-alpha（@yoborlon-alpha）开发并维护，当前版本 v2.0.8。
推荐 Skills