klaviyo-automation

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The required workflow relies on runtime tool calls to the Klaviyo toolkit (e.g., KLAVIYO_GET_CAMPAIGNS, KLAVIYO_GET_CAMPAIGN, KLAVIYO_GET_CAMPAIGN_MESSAGE) which return outsider-authored free text such as campaign/message HTML/text content that the agent will ingest into its LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly requires calling Rube MCP at runtime (it says "Always call RUBE_SEARCH_TOOLS first" and instructs adding https://rube.app/mcp as an MCP server), so remote tool schemas fetched from that URL can directly control the agent's toolset and prompts.