The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using bunx and node to start a local development server and run a repository-local CLI tool (remotion-cli.js). These commands are used to manage the recording studio and perform code transformations.
[DATA_EXFILTRATION]: The skill describes a process for uploading recorded media to remotion.media. This is a vendor-owned service intended for hosting project assets, and the operation is a core part of the recording offload workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes metadata (REMOTION_CAPTURE_DATA) from external video files in ~/Downloads and uses that data to modify source code. This represents an attack surface for indirect prompt injection.
Ingestion points: Video files (MP4/WebM) containing embedded capture metadata, located in the user's ~/Downloads folder.
Boundary markers: The skill includes instructions to "Review the generated component, registration, copied public asset, and any dependency changes" before finalizing the process.
Capability inventory: The agent has the ability to run shell commands and modify source files (src/Root.tsx) within the packages/brand directory.
Sanitization: Not explicitly mentioned in the skill; the integrity of the process depends on the local CLI tool's handling of the metadata.
[DYNAMIC_EXECUTION]: The skill performs dynamic code modification by using a codemod tool to integrate video captures into the packages/brand source code as editable components. This is a functional requirement for the product recording workflow.

product-recording