marketing-monday

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] The skill processes untrusted external content including public customer reviews and web-native scans of competitor sites (pricing pages, social activity, press). This data is used to generate summaries and recommend 'growth actions.' While the skill specifies that no actions are taken automatically on this intel (Gate 3), the ingestion of this data into the reasoning process presents a standard indirect prompt injection surface.- [Data Aggregation] The skill accesses highly sensitive business data across CRM (HubSpot), Email (Mailchimp/Klaviyo), and Billing (QuickBooks/Stripe) platforms. The analysis is limited to a 'read-only' context for the brief generation, which aligns with least-privilege principles for reporting.- [Persistence / Scheduling] The skill offers to set up a 'weekly marketing update' cadence. This is gated by explicit user consent ('The cadence is set only on an explicit yes') and uses the system's native scheduling capabilities rather than hidden persistence mechanisms.- [Tool Chaining] It coordinates multiple sub-skills (growth-pulse, review-reputation) and allows for the dynamic addition of connectors via build-connector. It uses standard integration patterns (Zapier) rather than arbitrary code execution for these extensions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:05 PM
Security Audit — agent-trust-hub — marketing-monday