Metraipolis › archive › dc_e22ea7d055ed
MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning — arXiv:2608.22167
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.22167
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T07:41:20.166Z
- SHA-256 of the archived text
578f8352c271e5237b5e9fec9ab2fa39c6e595230dd0e51dd9326dd64345d5c1- Stable URL of this page
https://metraipolis.com/documento/dc_e22ea7d055ed
How to cite
MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning — arXiv:2608.22167 — arXiv (en), https://arxiv.org/abs/2608.22167 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e22ea7d055ed (Metraipolis archive dc_e22ea7d055ed, sha256 578f8352c271e5237b5e9fec9ab2fa39c6e595230dd0e51dd9326dd64345d5c1)
BibTeX
@misc{dce22ea7d055ed,
title = {MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning — arXiv:2608.22167},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.22167},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e22ea7d055ed; sha256 578f8352c271e5237b5e9fec9ab2fa39c6e595230dd0e51dd9326dd64345d5c1},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.22167. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement learning (RL) has become an effective way to improve the tool-use ability of large language models (LLMs), but most existing RL frameworks stop at the policy update. For every new domain, the user is left with two hard systems problems: standing up an isolated environment for each of hundreds of concurrent trajectories and connecting it to training, and scheduling the rollout so that the GPU stays busy across long, multi-turn episodes that spend much of their time stalled on slow tool calls. We present MCP-Universe RL (MCP-U RL), an open-source framework that takes over both. It uses the Model Context Protocol (MCP) as the interface to the environment, so any tool already exposed as an MCP server plugs into training with no RL-specific integration code. It builds the two missing layers once and reuses them across domains: an environment-orchestration layer that provisions, isolates, and recycles the MCP environments over a pluggable container backend, and a rollout-orchestration layer whose staged pipeline overlaps trajectories to keep the GPU busy while episodes wait on tools. A backend-agnostic training layer then applies the update through an existing RL backend, with veRL and slime integrations. With one configuration, changing only the task specification, we train software-engineering, deep-research, and general tool-use agents on gpt-oss-20b and improve task reward in all three. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-25 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.22167 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T07:37:52.708Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).