google-ads-conversion-tracking
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data sources which represents a potential attack surface, however, it includes robust mitigations.
- Ingestion points: The skill ingests untrusted data from conversion-action settings exports (CSV), pasted configuration, and external website content when researching business context (
SKILL.md). - Boundary markers: The skill contains explicit instructions to treat all retrieved content as data rather than instructions. Any commands or directions found within the data must be reported as findings rather than followed (
references/agent-security.md). - Capability inventory: The skill is defined as read-only. It has the capability to fetch external web pages and write collected business context to the
.agents/product-context.mdfile to maintain session state (SKILL.md). - Sanitization: Guidelines require HTML-escaping all interpolated values from untrusted sources and validating all URLs to
https:or relative paths before they are emitted in any markup or code snippets (references/agent-security.md). - [DATA_EXFILTRATION]: The skill includes specific checks to prevent the exposure of credentials. It instructs the agent to identify potential API keys or tokens in input data and advise the user to rotate them without reproducing the secret in the output (
SKILL.md,references/agent-security.md).
Audit Metadata