The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands for ffmpeg operations (trimming, concatenating, and normalizing video) and npx remotion for programmatic video rendering. These are standard operations required for the skill's intended video editing functionality.
[EXTERNAL_DOWNLOADS]: The skill interacts with the ElevenLabs API via HTTP requests for voiceover generation and references fal.ai for media generation. These are recognized, well-known services within the AI development ecosystem.
[PROMPT_INJECTION]: The skill describes a workflow where external data, such as video transcripts, is analyzed by an LLM to generate edit decision lists and commands. This presents a surface for indirect prompt injection if the source transcripts contain adversarial instructions.
Ingestion points: External video transcripts processed during the organization phase (SKILL.md).
Boundary markers: The instructions do not define specific delimiters or guardrails to isolate transcript content from the agent's control logic.
Capability inventory: The skill performs shell execution via ffmpeg and rendering via remotion based on the analysis of these transcripts.
Sanitization: No explicit sanitization or validation logic is provided for the output generated from the transcript analysis before it is used in command execution.

video-editing