</>MCP Agents Market
MCP Server

deer-flow

by bytedance80.1kPythonUpdated 2026-08-17

An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.

Claude CodeCodexCursorWindsurf

DeerFlow is an open-source super agent harness MCP server that orchestrates sub-agents, sandboxes, memory systems, and extensible skills to handle long-horizon tasks spanning minutes to hours. Built by ByteDance, it enables AI agents to research, code, and create through a comprehensive framework supporting multiple LLM providers, tool integration, and isolated execution environments. The platform provides both a web interface and embedded Python client for running complex multi-step workflows with persistent context and safety controls.

Key Features

Multi-agent orchestration with sub-agent delegation and coordination across complex task hierarchies
Sandboxed code execution supporting Docker, AIO containers, and E2B environments with file system access controls
Long-term memory storage with session persistence and context engineering for extended workflows
Extensible skill system with customizable tools for web search, file operations, and domain-specific tasks
Support for multiple LLM providers including OpenAI, Anthropic, OpenRouter, vLLM, and reasoning models
Production-ready deployment with PostgreSQL/SQLite backends, Redis stream bridges, and multi-worker scaling
Interactive setup wizard and diagnostics toolkit for streamlined configuration and troubleshooting
Real-time tracing integration with LangSmith, Langfuse, and Monocle for observability

Use Cases

  • 01Deep research projects requiring iterative web searches, data synthesis, and multi-hour investigation workflows
  • 02Autonomous software development with code generation, testing, and debugging in isolated sandbox environments
  • 03Multi-step content creation combining research, writing, and artifact generation with persistent session context
  • 04Complex workflow automation requiring sub-agent specialization and coordinated task decomposition
  • 05Long-running data analysis pipelines with intermediate checkpoints and resumable execution
  • 06Prototyping agentic systems with the companion LLM Space desktop tool for step inspection and replay

Related MCP Servers

View more

deer-flow — FAQ

What is the DeerFlow MCP server?+

DeerFlow is an open-source super agent harness that coordinates sub-agents, memory systems, and sandboxed execution environments to complete complex tasks that can take minutes to hours. It provides a framework for building AI agents with persistent context, tool integration, and safety controls.

How do I install and set up DeerFlow?+

Clone the repository, run 'make setup' to launch the interactive configuration wizard, then start services with 'make docker-start' for Docker deployment or 'make dev' for local development. The wizard guides you through LLM provider selection, API key configuration, and sandbox mode settings in approximately two minutes.

Which AI clients work with DeerFlow?+

DeerFlow supports Claude Code, Codex, Cursor, Windsurf, and other coding agents through its one-line setup command. It also provides a web interface at localhost:2026, embedded Python client, terminal TUI, and optional LangGraph Studio integration for graph inspection.

What API keys and prerequisites does DeerFlow require?+

You need an API key for at least one supported LLM provider such as OpenAI, Anthropic, DeepSeek, or OpenRouter. Optional services include web search providers like Tavily, tracing platforms like LangSmith, and PostgreSQL/Redis for production deployments. The platform requires Python 3.12+, Node.js 22+, and Docker for sandbox-based execution.

Is DeerFlow free and open source?+

Yes, DeerFlow is released under the MIT license and freely available on GitHub. You are responsible for costs from third-party LLM providers, search APIs, and cloud infrastructure if deploying remotely.

What sandbox execution modes does DeerFlow support?+

DeerFlow offers Docker-based containers, AIO sandbox with provisioner support, and E2B remote environments for isolated code execution. Sandbox configuration includes file system access controls, bash command permissions, and resource limits for secure agent operation.

How do I install deer-flow?+

Open the source repository on GitHub and follow its README. deer-flow is a mcp server — MCP Agents Market links you directly to the official repo.

Is deer-flow free?+

deer-flow is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches