Security audit
Security checks for vulnerabilities and agentic risk
This Feishu skill is coherent, but it gives an agent broad workplace messaging and group-management power without strong confirmation or scoping guidance.
Install only if you are comfortable giving the agent Feishu permissions that can message users, create or modify chats, add members, and update persistent chat surfaces. Require explicit human approval for private messages, bulk sends, recalls, membership changes, and chat configuration updates.
Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
DELETE /open-apis/im/v1/messages/:message_id
实测心法：仅能撤回机器人自己在有效期内发送的消息。
The skill exposes broad messaging and group-management capabilities that can change workspace state and affect many recipients, but it does not prominently warn the operator to obtain confirmation before sending, recalling, adding members, or modifying chat configuration. In an agent setting, that omission increases the risk of unintended mass messaging, disruptive group changes, or privacy-impacting actions being executed without clear user awareness.
The documented P2P/private messaging capability enables direct outreach to users but provides no privacy or consent warning. In a multi-user workspace, an agent could use this to contact individuals unexpectedly, expose sensitive context to the wrong recipient, or bypass expected human review of outbound communications.
No suspicious patterns detected.