Unverified50% confidenceFactExact time
上下文工程是系统性地设计和优化输入给LLM的上下文信息，使模型在有限的上下文窗口内获取最相关的信息来完成任务
1
Sources
50%
Confidence
Long-term
Relevance
8/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified上下文工程是指系统性地设计、管理和优化输入给大语言模型的上下文信息的工程实践，与提示词工程侧重于单次交互的措辞优化不同79% similarVerifiedContext Engineering focuses on systematically building, managing, and optimizing all information fed to a large language model, including system prompts, conversation history, external knowledge bases, and user preference profiles.77% similarVerified上下文工程（Context Engineering）是一种新范式，为 AI 提供完整的体系化上下文，包括项目文档、代码示例、编码规则、开发模式和验证机制75% similarUnverifiedPrompt Engineering、tokens、上下文窗口和API调用是与大模型交互的基础前置知识73% similarUnverified上下文编译（Context Compilation）是在调用大模型之前对原始资料进行结构化处理的过程，包括扫描证据目录、生成证据清单、抽取哨兵字段、登记冲突风险四个步骤72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/824237API
curl https://kongchang.com/api/v1/knowledge/claims/824237MCP
get_claim(id=824237)