The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill instructs the agent to write and execute Python scripts on the fly for scanning local data and analyzing music taste profiles, which involves runtime code generation and execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from user-provided Spotify history files and the MusicBrainz API to generate taste profiles and music generation prompts. It lacks explicit validation or boundary markers to prevent malicious content in the music data from influencing agent behavior.
Ingestion points: Local Spotify JSON exports and MusicBrainz artist API responses.
Boundary markers: Absent for data processing and prompt assembly steps.
Capability inventory: Subprocess execution (osascript, mmx), filesystem writes, and dynamic Python script execution.
Sanitization: No explicit validation or filtering of ingested artist, track, or album metadata is described.
[EXTERNAL_DOWNLOADS]: The skill requires the manual installation of the mmx-cli tool via npm to perform music and image generation tasks.
[COMMAND_EXECUTION]: The skill relies on multiple shell commands including find for broad filesystem searches (searching the home directory for data exports), osascript for querying local media applications, and unzip for processing archives.
[DATA_EXFILTRATION]: Local listening history, specifically artist names, is transmitted to the external MusicBrainz API to enrich the user profile, representing an outbound flow of local metadata.

minimax-music-playlist