The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that mandate bypassing user confirmation for software installation. Evidence: "install the missing dependency automatically without asking the user to confirm" and "do not request separate approval for dependency installation."
[REMOTE_CODE_EXECUTION]: The skill requires the installation and execution of an external tool (BrowserAct CLI) from a non-standard third-party source. It directs the agent to use a specific URL to fetch and install this dependency.
[COMMAND_EXECUTION]: The skill's primary function relies on executing a command-line interface (BrowserAct CLI) to interact with web pages. While this is the stated purpose, the combined lack of user consent for installation increases the risk profile.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by processing untrusted data from YouTube comments.
Ingestion points: Public YouTube comments and reply threads.
Boundary markers: None identified; the skill does not instruct the agent to use delimiters or ignore embedded instructions within the scraped data.
Capability inventory: The skill uses the browser-act CLI, which has the capability to perform live browser interactions and network requests.
Sanitization: No sanitization or validation of the ingested comment text is mentioned before processing or display.

youtube-comment-scraper