Unverified50% confidenceFactExact time
将API文档直接嵌入提示词利用了大语言模型的In-Context Learning（上下文学习）能力，无需更新模型权重
1
Sources
50%
Confidence
Long-term
Relevance
9/10/2026
First Seen
Sources
Related Entities
Related Claims
Unverified大语言模型可通过工具调用（Function Calling）主动调用数据库查询和API接口，实现跨系统信息聚合，超越传统RPA工具的灵活性73% similarUnverified插件机制通过调用外部API让大语言模型获取实时数据、执行具体操作，解决LLM知识截止和无法与外部系统交互的局限72% 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72% similarUnverified将API文档下载步骤与计划生成步骤隔离是一种上下文卫生（Context Hygiene）实践，能确保上下文窗口中只有高质量的有效信息70% 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.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/887343API
curl https://kongchang.com/api/v1/knowledge/claims/887343MCP
get_claim(id=887343)