The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes several Python and Bash scripts (auto_grade.py, beat_grid.py, timeline_view.py, black_check.sh, loudness_check.sh, etc.) that execute ffmpeg and ffprobe for media processing. Commands are invoked using secure patterns (list-based arguments in Python) and strict input filtering via sed and grep in Bash to ensure path and parameter safety.
[EXTERNAL_DOWNLOADS]: The skill references standard dependencies and tools including pillow, numpy, librosa, scipy, and the gsap library. It uses npx to invoke the hyperframes CLI engine. All downloads and installations are directed towards official package registries (PyPI, NPM) or trusted vendor resources.
[INDIRECT_PROMPT_INJECTION]: The skill implements a sophisticated 'Gate' system (plan_gate.sh, scene_gate.sh) that parses project documentation (project.md). This represents a surface for ingesting untrusted data into the agent's logic; however, the skill mitigates this by using deterministic regex-based validation to enforce specific structural requirements before proceeding with assembly or rendering.
[CREDENTIALS_UNSAFE]: The skill correctly handles sensitive information by instructing that API keys (e.g., DASHSCOPE_API_KEY) be stored in environment variables or vendor-specific local configuration files, avoiding hardcoded secrets in the code or prompts.

qwen-mm-plugins-video-edit