The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites, creating a surface for indirect prompt injection where malicious instructions could be embedded in HTML tags or schema data. \n- Ingestion points: External HTML and SEO metadata retrieved via fetch_page.py and parse_html.py. \n- Boundary markers: The skill includes a dedicated security section instructing the agent to treat all fetched content as data, not instructions. \n- Capability inventory: The skill can execute local scripts, fetch external network data, and read/write to a local SQLite database. \n- Sanitization: Employs SSRF protection via google_auth.validate_url() and uses parameterized queries for database operations. \n- [COMMAND_EXECUTION]: Executes several internal Python scripts (drift_baseline.py, drift_compare.py, drift_history.py, and drift_report.py) through a local project-specific runner to process SEO data. \n- [EXTERNAL_DOWNLOADS]: Fetches content and performance metrics from user-supplied URLs to perform baselining and change detection.

seo-drift