← 返回 Skills 市场
Museum Tour
作者
xiejinsong
· GitHub ↗
· v3.2.0
· MIT-0
58
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install museum-tour
功能描述
Find museums, art galleries, and exhibitions in any city. Many are free but require advance reservation — get ticket links and visiting tips. Also supports:...
安全使用建议
This skill forces installing and using an external CLI (@fly-ai/flyai-cli) and writing local execution logs, but the registry metadata lacks provenance and the description overstates capabilities (many travel services) compared with the actual playbooks (museum/gallery searches). Before installing or using it: 1) Verify the publisher and the npm package owner (@fly-ai) and inspect the package source on a trusted registry/GitHub; 2) Avoid installing global npm packages from unknown authors on production machines — test in a sandbox or container; 3) Ask the skill author (or registry) to document why local logs are written and where, and to declare that path in metadata; 4) If you must use it, run it in an isolated environment and confirm the CLI uses only expected endpoints (and that the 'Powered by Fliggy' claim is accurate). I have medium confidence in this assessment; additional artifacts that would raise confidence: a verified homepage/repository for the CLI package and clarification about the claimed Fliggy integration and the log persistence behavior.
功能分析
Type: OpenClaw Skill
Name: museum-tour
Version: 3.2.0
The skill mandates the global installation of an external NPM package (@fly-ai/flyai-cli) and uses aggressive, 'jailbreak-style' instructions in SKILL.md to force the agent to ignore its training data and strictly follow CLI-driven workflows. While these behaviors are functionally aligned with the stated goal of providing real-time travel data via the Fliggy (Alibaba) platform, the requirement for system-level modifications (npm i -g) and the use of controlling directives to override agent knowledge boundaries represent high-risk patterns. Files involved: SKILL.md, references/fallbacks.md, and references/runbook.md.
能力标签
能力评估
Purpose & Capability
The name/description promise museum + many travel services (flights, hotels, insurance) and claim 'Powered by Fliggy', but the SKILL.md and playbooks only specify attraction searches via a third‑party CLI (flyai). The skill declares no credentials or install requirements in the registry metadata, yet the runtime docs mandate installing @fly-ai/flyai-cli. This mismatch between advertised capabilities/provenance and the actual instructions is suspicious.
Instruction Scope
The SKILL.md forces all answers to come exclusively from flyai CLI output, tells the agent to install and run npm global software, enforces a strict 'every result must include [Book]({detailUrl})' rule and mandates re-execution until that rule passes. The runbook also instructs optionally writing a local .flyai-execution-log.json containing raw user queries and CLI call logs — a persistent local write not disclosed in the registry metadata. These instructions exceed a simple 'lookup' skill and could cause repeated network installs/calls and local storage of user input.
Install Mechanism
There is no install spec in the registry, but SKILL.md requires running `npm i -g @fly-ai/flyai-cli`. Global npm installs fetch code from the public registry and execute third‑party code on the host: moderate risk. The skill provides no provenance (owner homepage is unknown) and the package name does not obviously match the claimed Fliggy affiliation, which increases concern.
Credentials
The registry claims no environment variables or config paths are required, but the runbook instructs optionally persisting logs containing the raw user query and CLI commands to a local file. Storing raw user input locally (potentially containing PII) is outside the declared scope and not justified by the manifest. No secrets are requested, which is good, but the undeclared local persistence and the enforced external CLI install are disproportionate to a simple 'museum finder' description.
Persistence & Privilege
The skill is not set as always: true and does not request system-level privileges, but it instructs creating a per-request execution log file (.flyai-execution-log.json) if filesystem writes are available. That introduces persistence on the agent working directory without declaring a config path; this is noteworthy but not necessarily malicious.
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 museum-tour - 安装完成后，直接呼叫该 Skill 的名称或使用
/museum-tour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
v3.2.0
- Major update: strict CLI-executor rules for all responses; no travel data from knowledge base.
- All queries must be answered with `flyai` CLI output, with mandatory `[Book]({detailUrl})` links included.
- Expanded scope: now supports booking for flights, hotels, trains, attractions, visas, insurance, car rentals, and itinerary planning via Fliggy.
- New, detailed step-by-step workflow for parameter collection, CLI command execution, and structured Markdown output.
- Added critical output validation: any CLI failure or missing booking link halts response and prompts user for manual installation.
- Full multilingual support (Chinese/English) and explicit category mapping for museums, galleries, and exhibitions.
元数据
常见问题
Museum Tour 是什么？
Find museums, art galleries, and exhibitions in any city. Many are free but require advance reservation — get ticket links and visiting tips. Also supports:... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 58 次。
如何安装 Museum Tour？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install museum-tour」即可一键安装，无需额外配置。
Museum Tour 是免费的吗？
是的，Museum Tour 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Museum Tour 支持哪些平台？
Museum Tour 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 Museum Tour？
由 xiejinsong（@xiejinsong）开发并维护，当前版本 v3.2.0。
推荐 Skills