The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes data from several untrusted external sources.\n
Ingestion points: Data enters the agent's context through nlp_analyze.py (analyzing external URLs/text), youtube_search.py (retrieving video comments and descriptions), gsc_query.py (fetching search queries), and pagespeed_check.py (analyzing web metrics).\n
Boundary markers: Output templates located in assets/templates/ utilize curly-brace placeholders but do not implement explicit delimiters or instructions to ignore embedded commands within that content.\n
Capability inventory: The skill can execute local Python scripts via the claude-seo run wrapper and generate file-based reports.\n
Sanitization: No explicit sanitization or filtering of external data is mentioned in the provided documentation.\n- [COMMAND_EXECUTION]: All primary features of the skill rely on the execution of local Python scripts (e.g., gsc_inspect.py, ga4_report.py) using the claude-seo run command.\n- [EXTERNAL_DOWNLOADS]: The documentation references the installation of the google-ads and google-analytics-data Python packages from standard registries. These are official packages for well-known services and are documented neutrally.

seo-google