Verified65% confidenceCautionExact time
MCP tools are essentially arbitrary code execution, indirect prompt injection is a real-world threat, and attack vectors can be hidden in resource content rather than in direct user input.
3
Sources
65%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
UnverifiedThe tool has a built-in function for pinning MCP tool definitions, and issues a warning when a connected MCP server subsequently modifies its tool definitions.80% similarUnverifiedIn MCP, tools are functions that the model can execute, resources are read-only data with no side effects, and prompts are reusable templates that users actively select.78% similarUnverified在MCP层可以注入恶意工具返回值、拦截审计Agent的工具调用、模拟工具失败等边界情况，使MCP层成为对抗测试的注入点与观测窗77% similarUnverifiedMCP tools typically require elevated system permissions to operate local software, and once maliciously exploited, the damage they can cause may far exceed that of vulnerabilities in ordinary open-source libraries.77% similarUnverifiedMCP 方案的核心理念是让人预先编写可控代码暴露给智能体，由人决定智能体能执行什么，而非放任智能体直连系统自由生成 SQL77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/547572API
curl https://kongchang.com/api/v1/knowledge/claims/547572MCP
get_claim(id=547572)