Unverified50% confidenceFactExact time
readOnlyHint and destructiveHint are metadata fields built into the MCP specification for communicating the degree of side effects of a tool to agent frameworks.
1
Sources
50%
Confidence
Long-term
Relevance
10/8/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedMCP 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.72% 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.71% 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67% similarUnverifiedHindsight provides APIs and SDKs, and supports MCP, allowing save, retrieval, and reflection functions to be exposed as Agent tools via MCP.67% similarUnverified该方案提供的MCP服务器包含一个read_document工具，用于解析文档且尤其擅长处理长文档66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/988963API
curl https://kongchang.com/api/v1/knowledge/claims/988963MCP
get_claim(id=988963)