DeerFlow
About this tool
ByteDance's open-source agent framework. Version 1.x focused on research; version 2.0 was rewritten for long-running tasks with sandboxes, long-term memory, subagents and skills. Tasks taking minutes to hours include search, reports, data analysis and presentations. Built-in research, systematic-review and paper-review skills work with MCP tools and messaging channels such as Feishu and WeCom.
- Feature 1: Built-in deep-research, systematic-literature-review and academic-paper-review skills.
- Feature 2: Local, Docker and Kubernetes sandbox options for code.
- Feature 3: Parallel subagents and long-term memory.
- Feature 4: stdio and HTTP MCP tools, a Gateway API and a Python client.
- Feature 5: Assign tasks through Feishu, WeCom or DingTalk.
Resource information
- Maintainer
- ByteDance; official company project under the bytedance GitHub organization.
- Type
- Standalone app
- Category
- Research agents & deep research
- Runs on
- Agent application
- Who it's for
- Research groups, research administrators and information analysts self-hosting research, systematic reviews, analysis and presentation workflows.
- License at time of listing
- MIT
- Source
- GitHub
