The Agent Skills Directory

[SAFE]: No high-severity security issues were detected. The skill's behavior is consistent with its stated purpose of assisting Japanese language learners.
[COMMAND_EXECUTION]: The scripts fetch_song.py and build_study.py execute external commands like curl, yt-dlp, and ffprobe. These are implemented using subprocess.run with list arguments rather than shell strings, which effectively prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill fetches song metadata (LRC) and audio (MP3) from tonzhon.com and music.163.com. These downloads are user-initiated and necessary for the primary functionality of generating study materials.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external music APIs. 1. Ingestion points: scripts/fetch_song.py reads LRC files from the web. 2. Boundary markers: No explicit delimiters are used in the prompt instructions. 3. Capability inventory: Subprocess execution is restricted to specific media tools. 4. Sanitization: The build_study.py script uses html.escape to sanitize all text before rendering it into the HTML study page, mitigating XSS and structural injection risks.

japanese-deep-translate