The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use curl to check for the existence of YouTube thumbnails (hq720.jpg vs hqdefault.jpg). This involves executing a shell command where parts of the command (the URL) are dynamically constructed based on external data.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes text from untrusted external sources and uses it to perform actions like file writing.
Ingestion points: Recipe content fetched from user-provided URLs, YouTube video descriptions, and external blog posts linked within those descriptions.
Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between the fetched data and the skill's own operational instructions.
Capability inventory: The skill can read local files, write new Markdown files to the vault, create creator 'stub' pages, perform network fetches via WebFetch/Playwright, and execute shell commands via curl.
Sanitization: There is no mention of sanitising or validating the external content (ingredients, instructions, or metadata) before it is processed or written to the file system.
[EXTERNAL_DOWNLOADS]: The skill frequently fetches data from external web services, including YouTube and various recipe websites, to extract information for the formatting process.

recipe-cleanup