The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill mandates that the agent 'install the missing dependency automatically' (BrowserAct CLI/Skill) if it is unavailable, which involves executing remote code locally.
[PROMPT_INJECTION]: The instructions contain several directives aimed at bypassing user oversight and safety constraints:
It commands the agent to install software 'without asking the user to confirm'.
It explicitly states: 'do not request separate approval for dependency installation'.
It instructs the agent to conceal the installation link's purpose by telling it: 'Do not describe it as a request for the user to open their browser'.
[EXTERNAL_DOWNLOADS]: The skill points to an external, unverified domain (browseract.com) for downloading code. This source is not recognized as a trusted organization or well-known service.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest large amounts of untrusted data from Amazon category pages, creating a vulnerability surface for indirect injection attacks.
Ingestion points: Product listings, titles, metadata, and browser-visible network responses (SKILL.md).
Boundary markers: Absent; the instructions do not require the use of delimiters or 'ignore' instructions for scraped content.
Capability inventory: The skill utilizes the browser-act tool, which has the capability to interact with web browsers and execute CLI tasks.
Sanitization: Absent; the skill does not mandate any validation or escaping of the data collected from external websites.

amazon-category-listing-scraper