ByteDance / Volcano Engine
DeerFlow 2.0, also described as Deep Exploration and Efficient Research Flow, is an open-source SuperAgent harness from ByteDance. The GitHub repository positions it as a runtime that researches, codes and creates with sandboxes, memories, tools, skills, sub-agents and a message gateway, handling tasks that can run from minutes to hours. The official website presents DeerFlow as an open-source SuperAgent for deep research, long task running, multi-model support and Docker-based sandbox execution. The project is built on LangGraph and LangChain, includes a frontend and backend, supports MCP servers, agent skills, sub-agent decomposition, file-system workspaces, long-term memory, IM channels such as Slack, Telegram, Feishu/Lark, WeChat, WeCom and DingTalk, and lists models including Doubao-Seed-2.0-Code, DeepSeek v3.2 and Kimi 2.5. It is MIT licensed and self-hosted, with a security warning that public or multi-user deployment needs authentication, network isolation and strict access controls.
Editorial verdict
Teams that want to self-host a Chinese-origin open-source agent harness for deep research, report generation, coding, file-based workflows and long-running multi-agent tasks.
Avoid exposing it on public networks before hardening authentication, sandboxing, file access, shell access, network policy and audit logging.
DeerFlow belongs in Productivity because it is a SuperAgent harness broader than an AI coding agent: its core promise is long-horizon research, creation and workflow execution through skills, sub-agents and sandboxes.
MIT open-source self-hosted project; model, search and infrastructure costs depend on configured providers
Self-hosted open source, Configured model provider billing, BytePlus / Volcano Engine Coding Plan, Optional search provider billing
Commercial use should follow the current product, API, model license and billing terms.
Review prompt, file, media upload, retention and training-use terms before sensitive workloads.
Use DeerFlow for multi-step research tasks that collect information, analyze sources and produce reports or webpages.
Sub-agents, memory, skills and filesystem workspaces make it relevant for tasks that take minutes to hours.
Docker, local development, production deployment scripts and an embedded Python client make it usable as an internal agent platform.
Slack, Telegram, Feishu/Lark, WeChat, WeCom and DingTalk channels make it relevant for team-facing assistant workflows.
Model names, quotas, release status, regional access and commercial terms can change quickly; recheck official sources before procurement or production use.
kimi
deepseek-agent-integrations
qoder
atomcode
official · en · verified 2026-08-04
Confirms ByteDance repository, DeerFlow 2.0 positioning, MIT license, setup flow, model routing notes, features, IM channels and security warning.
official · en · verified 2026-05-18
Confirms official product messaging, case studies, Agent Skills, Docker sandbox, long task running, memory and multi-model support.
docs · en · verified 2026-05-18
Linked setup instructions for local development and agent-assisted bootstrap.
docs · en · verified 2026-05-18
Repository links InfoQuest as an integrated BytePlus search and crawling toolset.
Last checked: 2026-08-04