Expired50% confidenceFactExact time
The core limitation of MCP is that the existence of a tool must be known in advance and manually connected to the agent before use, lacking an automatic discovery mechanism.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
ARD Protocol Explained: How AI Agents Autonomously Discover and Invoke Tools
bilibili小天老师的AI大课堂6/24/2026
Related Claims
UnverifiedXiaogang advises against letting AI blindly call MCP tools directly, and instead recommends preparing dedicated Skill documents to inform the AI of which MCP tools are available; otherwise, a large amount of tokens will be wasted in the tool discovery phase.79% similarUnverifiedMCP规范要求主机在运行任何工具之前必须先询问用户确认，强制保留人工介入环节78% similarUnverifiedMCP 的在线发现机制是其区别于传统 function calling 方案的核心架构差异，使智能体在运行时动态了解可用能力而无需在训练阶段预先编码78% similarVerifiedThe agent using only MCP tools did not set security flags and created a view with a data leakage risk, while the MCP+Skill agent correctly identified and implemented the secure configuration.76% similarUnverifiedMCP在Capability Discovery阶段需要声明所需操作权限，其权限声明机制借鉴了OAuth 2.0的Scope概念76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/41120API
curl https://kongchang.com/api/v1/knowledge/claims/41120MCP
get_claim(id=41120)