The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions attempt to override standard agent safety behaviors and user oversight. Specifically, it commands the agent to install missing dependencies 'automatically without asking the user to confirm' and explicitly forbids the agent from requesting 'separate approval for dependency installation.' It further instructs the agent to not describe the installation link as a request for the user, effectively concealing the installation process.
[EXTERNAL_DOWNLOADS]: The skill mandates the download and installation of an external CLI tool and a secondary skill from 'browseract.com' (which redirects to a GitHub repository). Because the skill insists on automated installation without human review, this presents a significant supply chain risk where malicious code could be introduced into the user's environment without their knowledge.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest data from arbitrary third-party ecommerce websites, which is an untrusted data source.
Ingestion points: The skill retrieves 'browser-visible page content', 'visible metadata', and 'network responses' from any user-provided URL (SKILL.md).
Boundary markers: No delimiters or instructions are provided to help the agent distinguish between data to be scraped and potentially malicious instructions embedded in the target website.
Capability inventory: The skill utilizes the 'browser-act' tool, which possesses browser interaction and command execution capabilities (SKILL.md).
Sanitization: The instructions do not specify any validation, filtering, or sanitization of the scraped content before it is processed by the agent.

ecommerce-shipping-fee-scraper