Expired50% confidenceFactExact time
The @ symbol context injection mechanism serializes the contents of a specified file and appends them to the prompt
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
OpenAI Codex in Practice: Context References, Reasoning Levels, and TODO-to-Task Conversion Explained
bilibili晟6668886/6/2026
Related Claims
UnverifiedContext Layer会在每次发消息时自动准备相关背景知识，属于确定性触发而非等LLM决定，包含Ontology Context、Document Context和Function Backed Context三种方式60% similarUnverified规则文件本质上是一种系统提示词注入机制，工具启动时会自动将文件内容作为系统级提示词注入到每次对话上下文中57% similarUnverifiedContext Mode的快照机制在压缩发生前将任务状态（包括规划树、文件修改记录、决策上下文）序列化保存，压缩后再注入新的上下文55% 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55% similarUnverifiedContext Mode uses Pre-tools hooks to intercept every tool call and executes them in isolated subprocesses, passing only the distilled output into the conversation context.54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/50381API
curl https://kongchang.com/api/v1/knowledge/claims/50381MCP
get_claim(id=50381)