The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute the tasteforge Python package, which is sourced from the Ito-Markets/ito-video GitHub repository. This repository is not on the trusted vendor list, and executing its CLI components represents a dependency on external, unverifiable code.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands including python3 -m tasteforge, ffmpeg, and ffprobe to process local media files and generate reports.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) as it processes user-influenced data from interview answers and workflow configurations which are then passed as arguments to local CLI tools.
Ingestion points: answers.json, media.json, workflow.json, and local style packs (SKILL.md).
Boundary markers: The skill lacks explicit LLM context delimiters for ingested data, although it emphasizes strict schema validation for the resulting JSON artifacts.
Capability inventory: The skill possesses the capability to execute shell commands (python3 -m tasteforge), run media processing tools (ffmpeg, ffprobe), and perform file system writes for EDL/FCPXML exports.
Sanitization: The skill documents extensive 'fail-closed' validation logic, requiring specific fields like provider_calls:0 and dry_run:true to be strictly enforced before any output is generated, which significantly mitigates the risk of unintended actions.

tasteforge-video