The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references an external site (browseract.com) for downloading a required CLI tool and agent skill.- [REMOTE_CODE_EXECUTION]: The instructions direct the agent to automatically download and install the BrowserAct dependency from a remote URL if it is not present on the system. This constitutes a download-and-execute pattern from an untrusted source.- [COMMAND_EXECUTION]: The instructions mandate that the agent automatically install dependencies "without asking the user to confirm" and explicitly forbid the agent from requesting "separate approval for dependency installation." This encourages the agent to bypass security guardrails that normally require user consent before executing installation commands or scripts from the internet.- [INDIRECT_PROMPT_INJECTION]: The skill captures and processes data from external websites (Yelp), which is inherently untrusted.
Ingestion points: Data is pulled from Yelp business pages and reviews via the browser-act tool.
Boundary markers: None are defined in the instructions to separate the scraped data from the agent's processing context.
Capability inventory: The agent is granted the capability to execute installation commands and interact with browser sessions.
Sanitization: The skill does not prescribe any sanitization for the captured data before it is processed or returned.

yelp-business-review-scraper