The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes several internal Python scripts (e.g., drift_baseline.py, drift_compare.py, and drift_report.py) via a local command runner to perform SEO audits, comparison logic, and report generation.
[SAFE]: The skill incorporates robust security controls, including SSRF prevention that blocks private IPs and metadata endpoints during URL fetching, and the use of parameterized SQLite queries to mitigate SQL injection risks.
[PROMPT_INJECTION]: The skill ingests untrusted data from external websites (HTML, meta tags, and JSON-LD schema) which serves as a potential surface for indirect prompt injection if an attacker-controlled site contains malicious instructions.
Ingestion points: External content is retrieved via scripts/fetch_page.py and processed by scripts/parse_html.py from user-specified URLs.
Boundary markers: Data is stored in a structured SQLite database (baselines.db), providing isolation from the execution instructions, though no specific prompt delimiters are mentioned for the final analysis output.
Capability inventory: The skill has the capability to execute local scripts, write reports to the filesystem, and perform network requests to external URLs.
Sanitization: The skill validates URLs for SSRF and uses SHA-256 hashing to verify content integrity, but it does not specify explicit filtering for the content of the parsed SEO elements.

seo-drift