drop-off-rescue

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE]: The skill accesses customer email addresses, tags, and event data (e.g., abandoned carts) from integrated marketing platforms (Klaviyo, Mailchimp, Rule, Get a Newsletter). This access is necessary for the skill's stated purpose of marketing automation.
  • [COMMAND_EXECUTION]: While 'Bash' and 'Write' are included in the allowed-tools configuration, the instructions do not utilize these tools for arbitrary shell commands or unauthorized file system modifications. The logic is primarily focused on API interactions with marketing services.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external marketing APIs and user-provided CSV files to draft email content.
  • Ingestion points: Data enters through list_profiles (Klaviyo), tool_list_members (Mailchimp), and other ESP-specific tools, as well as via direct CSV input in 'free mode'.
  • Boundary markers: No explicit delimiters are specified for the interpolated contact data.
  • Capability inventory: The skill has capabilities to read/write state to a context tree and interact with external marketing APIs to schedule communications.
  • Sanitization: The instructions do not define specific sanitization routines for contact names or CSV rows; however, the skill mandates a 'Confirmation gate' (Step 7), requiring a human user to review and approve all drafts and schedules before any external write operations occur, significantly mitigating the risk of automated exploitation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 06:43 AM
Security Audit — agent-trust-hub — drop-off-rescue