Seoul Subway
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: seoul-subway Version: 0.1.19 The skill is classified as benign. It transparently declares its use of a third-party proxy server (`vercel-proxy-henna-eight.vercel.app`) for all API calls and explicitly details the data sent (station names, search parameters, standard HTTP headers) and not sent (conversation history, personal data, credentials). The `SKILL.md` clearly states that `WebFetch` is used for network calls, not `curl` or other binary tools, mitigating shell injection risks. There are no instructions attempting prompt injection against the AI agent, nor any evidence of intent for data exfiltration, persistence, or unauthorized actions beyond its stated purpose of providing Seoul Subway information.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Your station names, route searches, IP address, and User-Agent may be visible to the proxy operator, which can reveal travel intent even though no account credentials are used.
The skill clearly discloses an external proxy data flow and identifies the query and header data the proxy receives.
This skill sends requests to a proxy server at `vercel-proxy-henna-eight.vercel.app` ... Station names ... Search parameters ... Standard HTTP headers (IP address, User-Agent)
Approve the proxy only if you are comfortable sharing those lookup details; avoid sensitive itinerary queries if that privacy tradeoff matters to you.
If you choose the persistent approval option, future subway lookups to that proxy may proceed without another domain prompt.
The skill asks the user to grant WebFetch access to its proxy domain and documents an optional persistent approval choice.
Recommended / 권장: Select `Yes` to allow access for this session ... You may also select `Yes, and don't ask again` for convenience, but only if you trust the proxy server.
Use session-only approval unless you trust the proxy and expect to use the skill regularly.