seoul-subway-arrival
Fail
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The instructions in
SKILL.mdrequire runningnpx -y @nomadamas/k-skill@0 instruct seoul-subway-arrival, which triggers a download of external code from the NPM registry. - [REMOTE_CODE_EXECUTION]: The use of
npxto fetch and execute a third-party package at runtime represents remote code execution from a non-whitelisted source. - [DATA_EXFILTRATION]: The skill uses
curlto transmit data tohttps://k-skill-proxy.nomadamas.org. While this is intended for querying transit data, it involves a non-whitelisted external domain and sends station names provided by the user. - [INDIRECT_PROMPT_INJECTION]:\n
- Ingestion points: Real-time arrival data is fetched from an external proxy in
instruction.mdand processed by the agent.\n - Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the external API response as untrusted content.\n
- Capability inventory: The skill includes shell command execution via
curlto interact with the API.\n - Sanitization: There is no evidence of filtering or validation of the data returned by the proxy before it is summarized for the user.
Recommendations
- AI detected serious security threats
Audit Metadata