deer-flow

Overview

DeerFlow is an open-source super agent harness that orchestrates sub-agents, memory, and sandboxes, powered by extensible skills. Version 2.0 is a ground-up rewrite that shares no code with v1; the original Deep Research framework is maintained on the 1.x branch. The project provides an official website with demos and allowlisted, read-only case studies that do not require sign-in. Setup begins by cloning the repository and running make setup, which launches an interactive wizard for choosing an LLM provider, optional web search, and execution/safety preferences such as sandbox mode, bash access, and file-write tools. The wizard generates a minimal config.yaml and writes keys to .env. Users can run make doctor to verify setup and get actionable fix hints. Operators can extend lead-agent, subagent, and DeerMem extraction prompts with literal prepend/append configuration without editing source templates. Optional per-model request_admission paces requests to help stay within provider request-per-minute limits. Administrators can open Settings → Models to add, edit, test, and enable/disable shared OpenAI-compatible Chat Completions models without editing config.yaml. DeerFlow has integrated the InfoQuest intelligent search and crawling toolset from BytePlus. Official DeepSeek models at https://api.deepseek.com or https://api.deepseek.com/v1 automatically use DeerFlow's DeepSeek adapter, preserving reasoning content across tool calls and honoring output token limits. YAML models remain read-only in this page and take precedence on name conflicts, while managed models are shared by the deployment. The encrypted catalog and a generated local encryption key are stored in $DEER_FLOW_HOME/managed-models/.