The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to manage the project structure and synchronize files.
Evidence: Step 10 uses mkdir -p and cp to create directories in the public folder and copy generated markdown files. The command parameters are derived from the path argument provided by the user.
[DATA_EXPOSURE]: The skill reads local project files to extract metadata and content.
Evidence: It accesses docs/week{N}/{slug}/eng/index.md and src/content/readings.ts. This access is necessary for the skill's primary function of content transformation and is limited to the local workspace.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external-facing markdown files (YouTube transcripts/translations).
Ingestion points: docs/week{N}/{slug}/eng/index.md and docs/week{N}/{slug}/kr/index.md.
Boundary markers: Absent (uses regex-based parsing).
Capability inventory: Modifies source code (readings.ts), creates files, and executes shell commands (mkdir, cp).
Sanitization: Relies on regex patterns for structural parsing, which provides a layer of validation against unstructured injection.

split-youtube-chapters