The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill communicates with https://k-skill-proxy.nomadamas.org to retrieve route summaries, geocoding, and reverse-geocoding data. This endpoint is a vendor-owned resource managed by the skill author.
[DATA_EXFILTRATION]: No sensitive local files (e.g., SSH keys, AWS credentials) are accessed. API credentials such as NAVER_MAP_CLIENT_ID and NAVER_MAP_CLIENT_SECRET are explicitly managed on the proxy server side, ensuring they are not exposed in the user's environment or prompt history.
[PROMPT_INJECTION]: The skill handles untrusted data through user-supplied addresses and coordinates, as well as JSON responses from the map proxy. While there are no explicit boundary markers to delimit external data, the skill lacks high-risk capabilities (such as dynamic code execution or file system writes) that would allow an indirect injection attack to escalate.
Ingestion points: User input from commands /route and /이동루트; JSON responses from the Naver Map proxy.
Boundary markers: Absent.
Capability inventory: No subprocess calls, exec/eval, file-write operations, or persistence mechanisms identified across scripts.
Sanitization: Employs standard URL encoding for parameters passed to the proxy server.
[COMMAND_EXECUTION]: The instructions direct the agent to use Python's standard libraries (urllib, argparse, json) or simple curl commands to interact with the vendor's API proxy. These operations are limited to the documented mapping functionality.

naver-map-route