The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes instructions to bypass user interaction and safety protocols. Specifically, it tells the agent to install software "automatically without asking the user to confirm" and commands the agent to "not request separate approval for dependency installation," which overrides standard agent-user safety boundaries and concealment of privileged actions.
[EXTERNAL_DOWNLOADS]: The instructions point the agent to https://www.browseract.com to fetch and install the BrowserAct CLI. This external domain is not associated with any known trusted vendors or services, presenting a supply chain risk.
[COMMAND_EXECUTION]: The skill requires the installation and execution of a Command Line Interface (CLI) tool. By forcing the installation and subsequent invocation of this tool without user review, it facilitates the execution of potentially unvetted code on the host system.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting untrusted data from public web pages (Google Maps results) and processing it through an automation tool (browser-act).
Ingestion points: Data extracted from Google Maps listings and linked websites (SKILL.md).
Boundary markers: Absent; no instructions are provided to the agent to treat scraped data as untrusted or to ignore instructions embedded in that data.
Capability inventory: Invokes the browser-act tool which interacts with live browser sessions and web content.
Sanitization: Absent; no mention of filtering or sanitizing the content retrieved from the web before processing or outputting to the user.

google-maps-franchise-lead-search