lark

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and parses user-generated Lark content via the Lark Blocks API (e.g., get_all_document_blocks and the /docx/v1/documents/{{document_id}}/blocks endpoint used by the /lark read <doc-id> command) and then uses that content as input to editing and messaging actions, so untrusted third-party document/wiki/chat content could influence agent decisions and tool use.