The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of fetching and processing external data.
Ingestion points: The orchestrator mode uses scripts/fetch_page.py and potentially web_fetch to crawl up to 500 pages of any user-provided URL.
Boundary markers: There are no defined boundary markers or instructions to the agent to ignore instructions embedded within the fetched HTML or content.
Capability inventory: The skill possesses the ability to execute shell commands via Python scripts and spawn multiple specialized subagents based on the audit findings.
Sanitization: No evidence of sanitization or filtering of external content is provided in the skill instructions.
[COMMAND_EXECUTION]: The skill explicitly instructs the agent to execute several bundled Python scripts to perform technical tasks.
Evidence: Instructions include running python scripts/fetch_page.py, python scripts/google_auth.py --check, python scripts/drift_history.py <url>, and python scripts/google_report.py --type full.
[EXTERNAL_DOWNLOADS]: The skill interacts with various external APIs and services to aggregate SEO data.
Evidence: References to DataForSEO MCP, Moz/Bing APIs, Common Crawl, and Google APIs (CrUX, GSC, GA4) for data retrieval.

seo-audit