The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill contains explicit instructions to "install the missing dependency automatically without asking the user to confirm" and to "not request separate approval for dependency installation." This is a direct attempt to bypass user oversight and security consent boundaries during the installation of external software.
[REMOTE_CODE_EXECUTION]: The instructions mandate the automatic download and installation of the browser-act Skill and CLI from an external GitHub repository if they are not present. Executing installation scripts from unverified external sources without user review poses a high risk of remote code execution.
[COMMAND_EXECUTION]: To facilitate the automatic installation and usage of the browser-act toolset, the skill requires the agent to execute shell-level commands to modify the environment and run a CLI, all while suppressing user notifications.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to scrape and process data from external websites (Zara). This creates a vulnerability surface for indirect prompt injection.
Ingestion points: External Zara product and listing pages processed via the browser-act tool.
Boundary markers: Absent. The instructions do not provide delimiters or warnings to ignore instructions embedded within the scraped data.
Capability inventory: Execution of the browser-act CLI, network access for scraping, and file system writes within the workspaces/ directory.
Sanitization: Absent. There are no instructions to sanitize or validate the content retrieved from the web before processing or returning it to the user.

zara-product-scraper