{"name":"Uni-Agent: A Scalable Framework for Training Long-Horizon AI Agents","description":"Uni-Agent is a powerful Python framework designed for training long-horizon agents at scale. It allows users to integrate existing agent harnesses, unify diverse agent tasks through an extensible interface, and run thousands of sessions concurrently for efficient data collection and training.","github":"https://github.com/verl-project/uni-agent","url":"https://osrepos.com/repo/verl-project-uni-agent","source":"osrepos.com","sourceDescription":"This repository profile is provided by osrepos.com, an open source repository discovery platform.","repositoryProfile":"https://osrepos.com/repo/verl-project-uni-agent","generatedFor":"open source discovery and AI-assisted research","markdown":"https://osrepos.com/repo/verl-project-uni-agent.md","json":"https://osrepos.com/repo/verl-project-uni-agent.json","topics":["Python","Reinforcement Learning","AI Agents","Machine Learning Framework","Scalable Training","Long-Horizon Agents","Distributed Systems"],"keywords":["Python","Reinforcement Learning","AI Agents","Machine Learning Framework","Scalable Training","Long-Horizon Agents","Distributed Systems"],"stars":null,"summary":"Uni-Agent is a powerful Python framework designed for training long-horizon agents at scale. It allows users to integrate existing agent harnesses, unify diverse agent tasks through an extensible interface, and run thousands of sessions concurrently for efficient data collection and training.","content":"## Introduction\n\nUni-Agent is a powerful Python framework designed for training long-horizon agents at scale. It allows you to bring any existing agent harness into reinforcement learning, unify diverse agent tasks through one extensible interface, and run agents concurrently at scale. The framework collects traceable trajectories as training-ready data for Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL).\n\n## Why Use It & Key Benefits\n\nUni-Agent offers several compelling features for developing and training advanced AI agents:\n\n*   **Plug in any agent harness:** Connect harnesses like Claude Code and Mini-SWE-Agent, or any harness compatible with OpenAI or Anthropic model endpoints, directly to the Uni-Agent Gateway.\n*   **Decouple agents, tasks, and infrastructure:** Build white-box agents from reusable `Agent`, `Tool`, `Task`, and `Sandbox` abstractions. Customize agent logic, tools, task environments, sandbox backends, and rewards independently while reusing the same evaluation and training runtime.\n*   **Run thousands of sessions concurrently:** Execute over 1,000 long-horizon, stateful sessions with distributed workers, pooled Gateway sessions, isolated sandboxes, and asynchronous scheduling. Every trajectory, log, and reward remains associated with the correct session for reliable evaluation, RL training, and data synthesis.\n*   **Reproducible training, verifiable results:** The project publishes runnable recipes with complete configurations, benchmark settings, result tables, and learning curves. These recipes provide tested starting points and make reported improvements easier to reproduce and verify.\n\nThe framework demonstrates strong performance in both parallel inference and verification, as well as agent reinforcement learning, with impressive scores on benchmarks like SWE-Bench and Terminal-Bench.\n\n## Installation\n\nTo get started with Uni-Agent, follow the official installation guide:\n\n*   [Install Uni-Agent](https://uni-agent.readthedocs.io/en/latest/quickstart/installation.html)\n\n## Examples\n\nUni-Agent provides a clear end-to-end path for development and training:\n\n*   [Launch a sandbox and run code](https://uni-agent.readthedocs.io/en/latest/quickstart/launch-sandbox.html) locally or with cloud services.\n*   [Run agent inference](https://uni-agent.readthedocs.io/en/latest/quickstart/agent-inference.html) at scale for benchmarking and trajectory generation.\n*   [Train an agent with RL](https://uni-agent.readthedocs.io/en/latest/quickstart/rl-training.html) with reproducible scripts and verifiable results.\n\nFor more detailed guides and examples, refer to the comprehensive [Uni-Agent documentation](https://uni-agent.readthedocs.io/en/latest/).\n\n## Links\n\n*   [Uni-Agent GitHub Repository](https://github.com/verl-project/uni-agent)\n*   [Uni-Agent Official Documentation](https://uni-agent.readthedocs.io/en/latest/)\n*   [Uni-Agent 26Q3 Roadmap](https://github.com/verl-project/uni-agent/issues/79)","metrics":{"detailViews":1,"githubClicks":1},"dates":{"published":null,"modified":"2026-09-30T23:59:02.000Z"}}