Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
目的地国家/地区旅行安全与禁忌提示
v1.0.0Generate a visually rich, card-based HTML info page covering city safety overview, local taboos & legal etiquette, emergency contacts, and transportation tip...
⭐ 0· 32·0 current·0 all-time
by@lmercy
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name, description, SKILL.md and template.html are coherent: the skill's goal is to generate a single-file HTML travel-safety card and the template plus placeholders match that purpose. There are no unrelated binaries, credentials, or config paths requested.
Instruction Scope
Most instructions are scoped to collecting user inputs, filling the bundled template, and obeying the listed content constraints. However, the 'Research content — Cross-check facts where possible' step implies web/API queries without specifying allowed sources or safety rules. More importantly, the workflow asks the agent to 'Save completed HTML as .html file and provide the link' but does not specify where to host that file (local attachment vs third-party upload). That open-ended step grants the agent broad discretion to call external hosting services or paste sites, which could result in unintended upload/exfiltration of user-provided context.
Install Mechanism
Instruction-only skill with no install spec, no external downloads, and no code execution beyond filling a local template. Low installation risk; nothing is written to disk by an installer.
Credentials
The skill requests no environment variables, credentials, or config paths. All required data are user-provided inputs or derived public information (e.g., emergency numbers, embassy info).
Persistence & Privilege
always:false and normal model invocation are set (no elevated always-on privilege). Autonomous invocation is allowed by platform default; combined with the open-ended 'provide the link' instruction this could broaden the blast radius (agent could autonomously research and then upload results). This is not inherently malicious but is a capability the user should consider.
What to consider before installing
This skill appears to do what it says (generate a travel-safety HTML card) and requests no credentials, but it leaves two operational details unspecified that you should confirm before installing or allowing autonomous runs:
- Where will the generated HTML be hosted/shared? The SKILL.md tells the agent to "provide the link" but does not constrain how. Ask or configure the agent to (a) return the HTML as an attachment or inline output instead of uploading to unknown third-party services, or (b) restrict uploads to approved internal storage. Prevent automatic uploads to paste sites, cloud storage, or other public endpoints unless you explicitly accept that.
- What sources will the agent use for "research" and fact-checking? If you care about accuracy and privacy, require the agent to cite official sources (government, local consulate, established travel advisories) and avoid pulling data from untrusted or user-contributed sites. You may also restrict or audit any web-browsing tool the agent uses.
Additional practical steps:
- Test the skill with a non-sensitive destination first and verify the generated HTML before any sharing.
- If you enable autonomous invocation, limit network/upload permissions or monitor the skill's activity so it cannot exfiltrate context unexpectedly.
- Review the final HTML for any embedded links or third-party resources before distributing it.
If you want, I can suggest a short modification to SKILL.md that constrains hosting behavior (e.g., "do not upload to external services; return file inline or attach to the conversation") and mandates citation of official sources — that would reduce the remaining concern.Like a lobster shell, security has layers — review code before you run it.
latestvk975fmjbsb4vppdy0zxndg0adn841r71
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
