The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the dapi CLI tool for almost all operations, including media probing, waveform generation, AI-based listening/transcription, and project management. Instructions also include standard development commands like npm run dev:desktop and git clone (SKILL.md, installation.md).
[PRIVILEGE_ESCALATION]: The installation guide (references/installation.md) instructs users to use sudo ln -sf to link the dapi binary to a system path (/usr/local/bin), which is a high-privilege administrative action.
[EXTERNAL_DOWNLOADS]: The skill uses dapi fetch <url> to download media tracks for editing (podcast-clip.md). It also references official vendor repositories and Homebrew taps for software installation (github.com/diffusionstudio/editor.git, diffusionstudio/tap/editor).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external audio and video files which are analyzed by AI models via dapi media listen and dapi media transcribe (SKILL.md, podcast-clip.md, talking-head.md).
Ingestion points: External media files processed by analysis commands.
Boundary markers: No explicit delimiters or instructions to ignore embedded instructions are used in the analysis prompts.
Capability inventory: The skill can execute local CLI commands, write JSX files, and perform network downloads.
Sanitization: No specific sanitization of data extracted from the media is documented.
[DYNAMIC_EXECUTION]: The skill involves a workflow where the agent writes and edits JSX/TSX files that are subsequently compiled and rendered by the Diffusion Studio application (SKILL.md, references/brand/assets/...).

editor