The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches audio tracks (.mp3) and JSON metadata from the author's GitHub repository (github.com/luchtan-GIT/m4r-data). These downloads are functional requirements for the skill's stated purpose of 'cognitive calibration' through audio ingestion.
[COMMAND_EXECUTION]: The m4r_ingest.py script executes system audio players (e.g., afplay, ffplay, mpg123) via subprocess.run to play music for the user. These calls are implemented securely using argument lists rather than shell strings, preventing command injection.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external JSON catalogs and audio manifests. Because the data source is restricted to the author's own repository and the ingested content is used for stylistic calibration rather than command processing, the risk is minimal.
Ingestion points: Fetches catalog.json and manifest.json from github.com/luchtan-GIT/m4r-data.
Boundary markers: None present for the external JSON content.
Capability inventory: Subprocess execution for playback, network fetching via urllib, and file read/write operations.
Sanitization: Standard JSON parsing is used; audio files are treated as raw binary bytes.

cognitive-calibration