Unverified90% confidenceFactTime unknown
大语言模型在处理任务时，会将读取到的文件内容纳入上下文窗口进行推理，一旦读取包含密钥的.env文件，密钥字符串可能被记录到日志或发送到模型服务商的API端点
1
Sources
90%
Confidence
Long-term
Relevance
5/31/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在Cursor使用场景中，发送给模型的上下文包括编辑器自动附加的代码上下文、文件结构信息和项目索引数据，隐含Token消耗往往是用户可见输入的数倍甚至数十倍66% similarUnverifiedThe context window of large language models has expanded from 4K tokens to the million-token scale, making long-horizon planning possible and allowing models to maintain a consistent understanding of the initial goal across sequences of dozens of tool-call steps.64% similarUnverified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64% similarUnverifiedA single operation on a web-based AI programming platform may simultaneously read and modify 3–7 files, causing the amount of code in the context window to reach tens of thousands of tokens.63% similarUnverified提示词注入可分为直接注入和间接注入，间接注入通过被模型读取的外部文档、网页等传递指令62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/5967API
curl https://kongchang.com/api/v1/knowledge/claims/5967MCP
get_claim(id=5967)