The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external platforms including Google Analytics reports, Search Console performance data, and BigQuery query results. This data could potentially contain malicious instructions that could influence the agent's behavior.
Ingestion points: Data is ingested via tools such as google_analytics_ga4_reports_run, google_search_console_performance_get, and bigquery_execute_query as described in SKILL.md.
Boundary markers: There are no explicit instructions for the agent to treat retrieval outputs as untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses state-changing capabilities such as creating or modifying tags in Google Tag Manager via google_tag_manager_tags_manage and executing SQL in BigQuery via bigquery_execute_query (SKILL.md).
Sanitization: No specific data validation or sanitization requirements are mentioned for handling external data before using it in subsequent operations.

analytics-insights