bytedance/deer-flow

Health v2 · maintenance only

Adoption not part of the score

84032 stars · 11713 forks observed · 2026-10-18

What it is AI-extracted, prompt v1, taxonomy v1, 2026-09-21, confidence 0.97

DeerFlow is an open‑source super‑agent harness that lets you compose, run, and manage complex AI agent workflows with memory, tools, and sandboxed execution. It provides both a core SDK (the Harness) and a reference application (the App) for building long‑horizon autonomous tasks.

Use cases

• automate multi‑step research projects that require web browsing, data analysis, and code generation
• build AI assistants that can remember context across long conversations and execute commands in a secure sandbox
• create custom LLM‑driven agents that integrate external tools such as search engines, databases, or code compilers
• orchestrate sub‑agents with specialized skills to solve large‑scale problems over minutes to hours
• prototype and benchmark new agentic workflows using the built‑in LLM Space desktop tool

When to choose

• you need a flexible, extensible framework for building long‑running, multi‑step AI agents
• your project requires memory persistence, tool usage, and sandboxed execution for safety
• you want tight integration with popular LLM ecosystems like LangChain, LangGraph, and BytePlus models
• you prefer a Python‑first SDK with Docker‑based sandbox support for reproducible runs

When to avoid

• you only need a simple single‑turn LLM call without orchestration or memory
• your environment cannot run Docker or lacks Python 3.12+ support
• you prefer a lightweight CLI tool rather than a full‑stack agent framework

Facets

framework · maturity active

workflow-automation artificial-intelligence large-language-models python windows super-agent harness subagents llm langchain langgraph deep-research sandbox-runtime agentic-framework sandbox memory tool-integration ai-agents docker linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="bytedance/deer-flow")