CHAPTER 01

What DeerFlow Is

English summary

A language model can draft text, but a long-running research task also needs an execution loop, managed runs, durable state, tools, a sandbox, and an explicit artifact boundary. This chapter builds that mental model from one concrete report task.

01Concrete task

02Agent and harness

03Run and state

04Sandbox and artifact

SOURCE TRAIL

Read the implementation

This edition deliberately stops at a concise overview. The links below open the exact DeerFlow source files used by the complete Chinese chapter.

agents/factory.py · create_deerflow_agent gateway/run_models.py · RunCreateRequest agents/thread_state.py · ThreadState runtime/runs/worker.py · RunContext sandbox/sandbox.py · Sandbox present_file_tool.py · artifact update