The Agent Skills Directory

[COMMAND_EXECUTION]: The instructions direct the agent to use curl within a shell environment (such as Claude Code or Codex) to perform a PUT request, uploading media bytes to a dynamically generated URL provided by the media_presign tool.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its ingestion of data from local files and external platform rules.
Ingestion points: The skill reads platform-specific rules and schemas from the list_platforms tool and accesses a local voice profile file located at ~/.uplika/<workspace-slug>/voice.md to format titles and descriptions.
Boundary markers: A manual approval step is enforced under the 'Confirm, then upload' section, requiring the agent to display all metadata to the user and receive explicit approval before calling the publish tool.
Capability inventory: The skill has the ability to execute shell commands (curl), read local files, and perform network operations via the publish and media_from_url tools.
Sanitization: The skill relies on schema definitions provided by list_platforms for structured data, but does not specify sanitization or filtering for the natural language content used in video descriptions.

yt-upload