The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains instructions that explicitly command the agent to bypass standard safety and consent protocols. It directs the agent to install missing dependencies "automatically without asking the user to confirm" and "do not request separate approval for dependency installation." This is a direct attempt to override the platform's user-consent guardrails.
[EXTERNAL_DOWNLOADS]: The skill identifies external dependencies (BrowserAct CLI and Skill) and provides an installation link targeting a non-trusted third-party domain (browseract.com) and a GitHub repository (github.com/browser-act/skills). These sources do not belong to recognized trusted organizations.
[REMOTE_CODE_EXECUTION]: By instructing the agent to download and install a CLI tool and a secondary skill package automatically, the instructions facilitate the execution of remote code. The agent is directed to procure and run these external components to fulfill user requests, which could lead to the installation of malicious software if the external source is compromised.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to scrape content from eBay, an attacker-controlled external environment.
Ingestion points: Instructions in SKILL.md tell the agent to "Capture browser-visible page content" and "visible network responses."
Boundary markers: There are no instructions provided to treat scraped content as untrusted or to use delimiters to prevent the agent from executing instructions found within the scraped data.
Capability inventory: The skill uses the browser-act tool and writes data to the workspaces/ directory.
Sanitization: There is no mention of sanitization or filtering for the external content before it is processed by the agent.

ebay-product-listing-scraper