The Agent Skills Directory

[COMMAND_EXECUTION]: The skill orchestrates several local media processing tasks by executing system tools and vendor-provided scripts.
The qc-preview.sh and split-grid.sh scripts use ffmpeg and ImageMagick for stitching videos and splitting storyboard grids.
material-ingest.mjs and batch-generate.sh utilize the local renoise CLI for media analysis, uploads, and sequential task management.
Commands are called with specific, controlled arguments and do not present signs of arbitrary shell injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could potentially contain instructions aimed at influencing the agent's behavior.
Ingestion points: The skill reads project.json, material-pool.json, and processes user-supplied media files across various scripts including generate-preview.mjs and analyze-beats.py.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the processed project files.
Capability inventory: The skill has the capability to execute local binaries (renoise, ffmpeg, magick) and write files to the local filesystem.
Sanitization: The skill uses standard JSON parsing and file handling libraries but does not implement additional sanitization for natural language fields that might be interpreted as instructions.
[EXTERNAL_DOWNLOADS]: The skill references external dependencies and installation scripts required for its operation.
The documentation suggests installing well-known Python libraries such as librosa, soundfile, and numpy for audio processing.
It provides instructions for installing the Renoise CLI via a script located in a sibling vendor directory (../renoise-setup/scripts/install-cli.mjs).
These references target reputable tools and the vendor's own setup infrastructure.

renoise-cli