The Agent Skills Directory

[SAFE]: The skill is a diagnostic tool designed to perform a one-time audit of Google Analytics 4 properties. It utilizes a specific MCP (Model Context Protocol) server to interface with the GA4 API.
[SAFE]: The instructions explicitly emphasize a read-only approach ('Never change property settings'), ensuring that the skill does not inadvertently modify sensitive analytics configurations.
[SAFE]: The skill includes a proactive check for Personally Identifiable Information (PII) within GA4 event parameters, helping users identify compliance risks.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface: The skill ingests data from GA4 API responses which could contain attacker-controlled strings (e.g., malicious event or parameter names) designed to influence the agent's output.
Ingestion points: Event and parameter names fetched via tool_get_metadata, tool_run_report, and tool_list_conversion_events in SKILL.md.
Boundary markers: Absent; the skill does not explicitly use delimiters to wrap the ingested strings when generating the final report.
Capability inventory: The skill has the capability to write findings via mcp__cogny__create_finding and uses WebFetch/WebSearch for external lookups.
Sanitization: There is no evidence of sanitization or escaping applied to strings retrieved from the GA4 property before they are processed for the final audit report.

ga4-audit