korean-middle-korean
Fail
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the execution of
npx -y @nomadamas/k-skill@0, which downloads code from the NPM registry at runtime. - [REMOTE_CODE_EXECUTION]: By instructing the agent to run
npxwith a remote package to obtain 'primary instructions', the skill allows an external, untrusted source to dynamically define the agent's behavior and execute code on the host system. - [COMMAND_EXECUTION]: The helper script
scripts/korean_middle_korean.jsusesfs.readFileSyncto read from arbitrary file paths provided via the--fileargument. This capability could be exploited by an agent to expose sensitive local files if prompted to process them. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user text and file contents as input for the conversion process. The absence of specific boundary markers or instruction sanitization for the converted output creates a surface for indirect prompt injection if the agent later interprets the converted text as instructions.
Recommendations
- AI detected serious security threats
Audit Metadata