Unverified50% confidenceFactExact time
科研自动化 Agent 通常基于 LLM 的工具调用与任务规划能力，结合代码执行沙箱、文件系统访问、学术检索及 GitHub/Hugging Face API 等外部工具，形成感知-规划-执行-反馈的闭环
1
Sources
50%
Confidence
Long-term
Relevance
7/13/2026
First Seen
Sources
科研自动化Agent实测：零样本复现NeurIPS顶会论文全流程
bilibiliNitres7/9/2026
Related Claims
UnverifiedAgent型助手能够接收高层意图指令，自主规划步骤、调用工具、读写文件、运行测试，完成跨越多文件乃至整个代码库的复杂任务79% similarUnverified在Agent架构中，LLM负责意图理解和步骤规划，实际执行委托给计算器、搜索引擎、API端点等外部工具77% similarUnverifiedAI Agent在LLM基础上增加感知-推理-行动循环，能调用日历API、邮件客户端、文档编辑器等外部工具完成操作76% similarUnverifiedAgentic 编程工具能够理解整个任务目标，自主制定执行计划，并在执行过程中根据反馈动态调整策略76% similarUnverifiedSince 2023, projects like Stanford's Generative Agents, Microsoft's AutoGen, and Andrew Ng's Agentic Workflow have demonstrated that LLM-based agents can handle complex cognitive tasks including planning, execution, and reflection76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/500144API
curl https://kongchang.com/api/v1/knowledge/claims/500144MCP
get_claim(id=500144)