k-skill-korean-ai-tools

Pass

Audited by Gen Agent Trust Hub on Mar 27, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires installing several automation packages and CLI tools, such as k-skill, kakaocli, and hwp.js, from the public npm registry.
  • [COMMAND_EXECUTION]: Multiple shell commands are utilized for environment configuration and automation tasks, including k-skill-setup, sops, and various service-specific CLI operations for SRT and HWP conversion.
  • [CREDENTIALS_UNSAFE]: The skill manages authentication data for train reservations and external API keys. It recommends using environment variables and sops encryption for secure storage, which is a positive security practice.
  • [DATA_EXFILTRATION]: The skill accesses sensitive user information, including KakaoTalk conversation history and train reservation details, to perform automated service tasks.
  • [PROMPT_INJECTION]: There is a risk of indirect prompt injection due to the skill's processing of untrusted external data.
  • Ingestion points: Processes HWP documents via HWPProcessor and retrieves live data from sports (KBO) and subway APIs.
  • Boundary markers: Absent; the skill does not use specific delimiters to isolate processed data from instructions.
  • Capability inventory: The agent can send messages, perform reservations, and execute various CLI commands.
  • Sanitization: External content is processed without visible validation or sanitization steps in the examples provided.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 27, 2026, 10:23 AM