access

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

• Third-party content exposure detected (high risk: 0.85). The skill reads and acts on pending pairing entries written by the Feishu channel server (~/.claude/channels/feishu/access.json → pending), which originate from untrusted Feishu messages/user-generated input and are used to approve/add users (e.g., the pair <code> flow), so third-party content can directly influence actions.