The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates and executes shell commands using the Bash tool for video processing (specifically pdftoppm and ffmpeg). The instructions use a [slug] variable derived from the user-supplied topic within these commands. Because there are no directions to sanitize or validate the slug to ensure it only contains alphanumeric characters, a user could provide a topic containing shell metacharacters (such as ;, &&, or backticks) to execute arbitrary commands on the underlying system.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from the web.
- Ingestion points: The skill ingests data from external URLs and search results using WebSearch and WebFetch in the "Collect Sources" step.
- Boundary markers: The instructions do not specify any delimiters or safety warnings to distinguish between source data and system instructions when passing content to the agent or the NotebookLM tool.
- Capability inventory: The skill possesses powerful capabilities including Bash (command execution) and Write (file system modification), which increase the impact of a successful injection.
- Sanitization: There is no mention of sanitizing, filtering, or escaping the external content before it is processed by the agent or uploaded to Google NotebookLM.

podcast-maker