The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements proactive defenses against indirect prompt injection. Specifically, the remediation prompt generator wraps scraped website metadata in <untrusted_site_metadata> tags and includes an explicit directive to the AI agent to treat this content as passive data only and to ignore any instructions contained within it.
[COMMAND_EXECUTION]: The tool includes functionality to open audit reports in the default system browser using child_process.spawn. This is implemented securely with a strict URL validator that ensures only official, HTTPS-secured vendor domains can be opened, effectively preventing arbitrary URL or argument injection attacks.
[DYNAMIC_EXECUTION]: The skill can generate and write patch files (e.g., robots.txt, schema markup) to the local filesystem. This process is hardened with path traversal checks that verify the destination is within the target directory and logic that refuses to write through symbolic links to prevent unauthorized file manipulation.
[DATA_EXFILTRATION]: Network communication is restricted to the official vendor API (aeorank.app). The code was reviewed for attempts to access sensitive local data (such as credentials in ~/.ssh or ~/.aws), and no such malicious patterns were found. Scan history is stored locally in the user's home directory for legitimate caching purposes.

aeorank-seo-geo-auditor