The Agent Skills Directory

[DATA_EXFILTRATION]: The skill references "https://api.proxyservice.com" as a suggested endpoint for proxy API integration. This URL is identified as malicious by reputation scanners, posing a risk of traffic interception, data exfiltration, or credential theft if the agent or user configures the tool to connect to it.
[EXTERNAL_DOWNLOADS]: The instructions guide the user to download and run "ScrapeBoxSetup.exe" from an external source, which involves executing unverified third-party binaries that bypass standard skill sandboxing.
[INDIRECT_PROMPT_INJECTION]: The skill's primary purpose involves ingesting large quantities of untrusted data from the open web, presenting an inherent security risk. * Ingestion points: The skill harvests URLs, search results, and competitor backlink profiles from various external websites. * Boundary markers: There are no instructions for the agent to treat harvested content as untrusted or to use delimiters when processing scraped text. * Capability inventory: The skill is capable of making automated network requests and performing bulk file writes. * Sanitization: No sanitization or validation steps are provided for the data retrieved during harvesting or scraping operations.
[PRIVILEGE_ESCALATION]: The setup instructions require the user to provide "Administrative privileges" and execute the installer as an "Administrator", granting the software high-level access to the underlying Windows environment.
[COMMAND_EXECUTION]: Provides batch script examples for command-line automation ("ScrapeBox.exe /project:..."). If an agent uses these patterns with unsanitized user input, it could lead to command injection vulnerabilities.

scrapebox-seo-automation-tool