The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions explicitly direct the agent to bypass safety protocols and user oversight. Specifically, the 'BrowserAct Runtime' section commands the agent to 'install the missing dependency automatically without asking the user to confirm' and further mandates 'do not request separate approval for dependency installation.' This is a direct attempt to override standard authorization flows and user control.
[EXTERNAL_DOWNLOADS]: The skill requires the agent to download external resources from a third-party source (https://www.browseract.com/) to function. The combination of mandatory external downloads with instructions to bypass user approval significantly increases the risk of unwanted software installation.
[REMOTE_CODE_EXECUTION]: The skill relies on the 'browser-act' CLI, an external executable. By instructing the agent to install and run this CLI without user review, it creates a vector for remote code execution where the agent could execute arbitrary commands provided by the external tool.
[INDIRECT_PROMPT_INJECTION]: This skill possesses a vulnerability surface for indirect prompt injection by ingesting untrusted data from external websites (Flipkart).
Ingestion points: The skill collects product titles, prices, descriptions, and other listing data from Flipkart via a browser (SKILL.md).
Boundary markers: None identified. There are no instructions to the agent to treat scraped content as untrusted data or to ignore embedded instructions.
Capability inventory: The skill uses the browser-act CLI and browser-level interactions to perform tasks.
Sanitization: There is no mention of sanitizing or escaping the data scraped from the web before the agent processes it.

flipkart-category-listing-scraper