Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

go-now

v1.0.0

你的私人旅行搭子,帮你弄清楚想去哪里、哪里适合你、然后推你一把立刻行动。当用户说"我想出去玩"、"周末不知道干嘛"、"好累想放松"、"想去旅行但不知道去哪"、"最近压力好大想逃离"、"有假期不知道怎么安排"、"想出去走走"、"不知道去哪玩"等模糊的旅行意愿时触发。也适用于用户表达了拖延、犹豫、选择困难等旅行决策障...

0· 62·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
high confidence
Purpose & Capability
Overall the requested capabilities (finding destinations, producing an HTML travel poster, helping user decide and book) match the skill description. However, the SKILL.md instructs network-based IP geolocation and local file writes/auto-opening of a generated HTML file — these are plausible for a travel assistant but are not declared in the manifest (no required binaries, no config paths) and therefore deserve explicit user consent and/or declaration.
!
Instruction Scope
Instructions direct the agent to: (1) automatically obtain the user's city via public IP geolocation APIs (curl to ipinfo.io, ip-api.com, myip.ipip.net) without requiring explicit user consent; (2) run shell curl commands (curl is used but not declared as a required binary); (3) collect personal details (name, avatar URL, vehicle plate, etc.) during conversation and embed them in output; (4) save a generated HTML file to ~/Desktop/go-now-{destination}-{date}.html and automatically open it in a browser. These actions involve network calls, potential privacy-sensitive data collection, filesystem writes, and remote-resource loading (images/fonts). The SKILL.md also requires a self-contained HTML file but the provided template imports Google Fonts (external network call), an internal inconsistency.
Install Mechanism
This is an instruction-only skill with no install specification or code to write to disk, which is low risk from an installation perspective. However, runtime instructions call external services and expect shell utilities (curl) to be present despite no binaries declared.
!
Credentials
The skill requests in-conversation personal information (name, avatar, vehicle info, travel preferences) which is reasonable for personalization, but the instructions also embed remote resources (user-provided avatar URLs and the template's Google Fonts import). Embedding remote images or fonts can cause the user's client or the agent environment to fetch external resources, leaking data (e.g., that the user generated a file) or exposing IP addresses to third parties. No credentials or env vars are declared, which is good, but the implied network access and potential for exfiltration are disproportionate without explicit consent and safeguards.
Persistence & Privilege
The skill does not request persistent platform privileges (always: false). It does instruct the agent to create a file on the user's Desktop and open it in a browser, which is a one-off persistence action rather than ongoing background privilege. Still, writing to a user-visible filesystem path and auto-opening a file should be gated by explicit user permission in the conversation.
What to consider before installing
This skill appears to be a genuine travel assistant, but it performs several actions that have privacy and operational implications. Before installing or enabling it, consider the following: - Consent for location: It automatically attempts IP-based location lookups using public APIs (ipinfo.io, ip-api.com, myip.ipip.net) at conversation start. Ask the skill to get explicit permission before doing any geolocation (or prefer asking the user to provide city). - Shell/network use: The runtime expects to run curl commands and make outbound HTTP requests. Ensure you are comfortable with the agent making network calls to third-party endpoints, and verify whether your environment allows/monitors such calls. - Filesystem writes & auto-open: The skill will save an HTML file to ~/Desktop and open it in your browser. Confirm in-chat before any file is written or opened, and review the generated file before sharing. - Remote resources & data leakage: The demo template imports Google Fonts and allows embedding avatar URLs/images. Remote fonts/images trigger outbound requests when the file or preview is loaded, which can leak a URL load to external hosts. Prefer inlining assets or disallow remote image embedding, and do not supply sensitive images/URLs (e.g., private avatars or images behind authentication) unless you trust the behavior. - Inconsistencies to fix: The SKILL.md claims the HTML should be self-contained, but the provided template imports fonts from Google Fonts; and curl usage is assumed but not declared. Ask the author to (a) declare required binaries and network endpoints, (b) make the HTML truly self-contained or explicitly list external resources, and (c) require explicit user consent before geolocation, personal data collection, saving files, or opening browsers. If these issues are acceptable and you trust the skill author, you can proceed; otherwise request the fixes above or restrict the skill's ability to perform network calls and filesystem writes.

Like a lobster shell, security has layers — review code before you run it.

latestvk970t057w1d1pksvktk56x71z1843w6v

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments