intercom-automation

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflows (e.g., "Manage Conversations" and "Reply and Manage Conversation State") explicitly call Intercom APIs like INTERCOM_GET_CONVERSATION / INTERCOM_LIST_CONVERSATIONS that ingest user-generated conversation bodies and contact data (conversation bodies support HTML, noted under "Known Pitfalls"), so untrusted third-party content is read and can materially influence replies, assignments, and other agent actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires adding and calling the Rube MCP endpoint (https://rube.app/mcp) at runtime and explicitly instructs using RUBE_SEARCH_TOOLS to fetch current tool schemas, which directly control the agent's available prompts/actions, so this external URL is a required runtime dependency that can change agent instructions.