jd-opensource

OxyGent: Build Modular Multi-Agent Systems in Python

OxyGent is a Python framework that uses its Oxy abstraction to assemble tools, models, and agents into observable, evolvable multi-agent systems. It is aimed at developers building and running agent teams with flexible collaboration patterns.

Stars
2.1k
Forks
276
Stars, last 30 days
+3
Last commit
1 day ago
Contributors
26

Our take

Maintaineda few commits in the last 90 days

Health score

46/100

How it is scored
Activity
30
Community
59
Issues
53
Pull requests
43

+3 stars in the last 30 days

OxyGent is a focused Python framework for modular agent teams, with recent maintenance activity and a broad contributor base, but no release in the past year and limited recent issue and pull-request activity.

Good fit if

  • You want to define multi-agent systems in Python using OxyGent's component model.
  • Your project can use an API-compatible LLM endpoint and you are comfortable supplying its credentials.
  • You want to evaluate a framework whose README describes dynamic planning, evaluation, and distributed scheduling.

Look elsewhere if

  • You need a recent versioned release as a prerequisite for adoption.
  • Your deployment depends on supplied Docker, Compose, or Helm configuration, which the repository requirements do not indicate.
  • You need a project with active recent issue resolution: the measured data shows only a small amount of issue and pull-request handling recently.
All health signals
Last commit2026-10-27 (1 day ago)
Commits, last 90 days3
Releases, last 12 months0
Contributors26 (top contributor: 54% of commits)
Issues closed, last 90 days2
Pull requests merged, last 90 days1
Project age16 months

Checked on 2026-10-29 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

2,077

Forks

279

Change since Nov 17, 2026: +1 stars·+1 forks

Overview

OxyGent is a Python framework for building multi-agent systems around its Oxy abstraction, which standardizes tools, models, and agents as modular components. It aims to make agent systems easier to assemble, inspect, and evolve than rigid workflows, while supporting different agent topologies and dynamic planning.

The project is for developers who want to compose agents and tools through Python interfaces, then run them as a coordinated system. Its README also describes evaluation and distributed scheduling capabilities, but teams should validate those against their deployment needs and the current documentation.

Key Features

  • Compose tools, models, and agents as reusable Oxy components.
  • Define agent teams in Python, including a master agent with sub-agents.
  • Use ReAct agents and the dynamic or hybrid planning patterns described by the project.
  • Configure an LLM through an API-compatible endpoint and environment variables.
  • Run a multi-agent system through the MAS interface and its web service.
  • Connect to MCP tools, with Node.js required when using MCP.
  • Use built-in evaluation and visualization capabilities described in the README to inspect and improve agent systems.

Use Cases

  • Build an internal assistant that delegates time, file, or calculation tasks to specialized agents.
  • Prototype a multi-step task workflow where agents need to decompose work and adapt as it proceeds.
  • Assemble reusable agent and tool components for several related business scenarios.
  • Explore a Python-based multi-agent architecture before deciding whether its planning and deployment approach suits production requirements.

What you need

Detected in the repository

  • Python (from requirements.txt)
  • A test suite and automated checks on GitHub Actions

Getting Started

Use Python 3.10 or newer and configure an API-compatible LLM endpoint:

pip install oxygent

See the README and official documentation for a complete example and configuration details.

License in Plain Words

No standard license was detected. Without a license, the default copyright rules apply: you may read the code, but not reuse it. Check the repository or ask the authors.

Alternatives

  • autogen: AutoGen centers on conversational agent orchestration and is in maintenance mode, while OxyGent focuses on composing agent teams through its Oxy abstraction.
  • Agentarium: Agentarium emphasizes interactive environments, agent memory, actions, and checkpoints, while OxyGent focuses on assembling tools, models, and agents into observable systems.
  • ROMA: ROMA uses recursive planning to decompose complex tasks among agents, while OxyGent provides a broader framework for composing teams with flexible collaboration patterns.
  • graphbit: GraphBit uses a Rust core for concurrent, reliable workflow execution, while OxyGent emphasizes its Oxy abstraction for assembling and evolving agent systems.
ProjectLanguageLicenseStarsStatus
OxyGentPython–2.1kMaintained
autogenPythonCC-BY-4.061.3kQuiet
AgentariumPythonApache-2.0935Maintained
ROMAPython–5.2kQuiet
graphbitRustApache-2.0585Quiet

Considerations

  • The measured health data marks the project as maintained, with a few commits recently, but reports no releases in the past year. Plan to assess the current code and package version rather than relying on a recent tagged release.
  • The project has contributions from multiple people and is not flagged as concentrated around one contributor. Recent issue closures and merged pull requests are limited, so do not assume fast support or review turnaround.
  • Python 3.10 or newer and an LLM provider API key are needed for the documented setup. Node.js is an additional requirement when using MCP.
  • The README identifies the license as Apache 2.0, while the repository metadata reports NOASSERTION. Confirm the applicable license in the repository before adopting or redistributing the code.

Found this useful?

Share it with someone who would like OxyGent.

Comparisons