mlsanityData to 14 Oct 2026

MCPDisCo: Seesaw-Mitigated Tuning for MCP-Native Agents

arXiv 2610.14956 ↗First posted 10 Oct 2026cs.AIcited by 0

On the map: LLM agents and tool use → Reinforcement learning for LLM agents also fits: General LLM agent frameworks and benchmarks

Abstract (arXiv)

The Model Context Protocol provides a shared interaction substrate for the agentic web, allowing heterogeneous services to expose capabilities through a standardized interface. Learning to use this ecosystem entails a hierarchy of capabilities: tool-format skills, server-level transfer, and cross-server composition. We present MCPDisCo, an ecosystem-scale reinforcement learning framework for MCP-native agents. Using a Thought-Markov Decision Process (Thought-MDP), the framework explicitly represents internal reasoning alongside routing, execution, and termination and trains on real-server, long-horizon interactions spanning 70 MCP servers and 527 tools across six domains. We study the reasoning--tool seesaw, a joint-learning failure mode in which improvement in one objective can accompany deterioration in the other. To mitigate potential interference, MCPDisCo combines token-masked reward attribution with bidirectional gradient surgery: the former separates direct supervision for reasoning and action spans, while the latter removes conflicting components from both objective gradients in shared parameters. MCPDisCo achieves higher success with fewer chain-of-thought tokens. On the out-of-distribution MCP-Atlas and the GAIA benchmark, MCPDisCo maintains monotone gains, confirming that the learned behavior generalizes beyond the training server distribution and broader agentic tasks. Model weight and implementation are available.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use. Builds on 19 of its 23 references that are in the map. First posted on arXiv 10 Oct 2026.

Builds on

19 of 23 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning23 Aug
  2. 2026Scaling Agentic Capabilities, Not Context: Efficient Reinforcement Finetuning for Large Toolspaces5 Mar
  3. 2026Reasoning and Tool-use Compete in Agentic RL:From Quantifying Interference to Disentangled Tuning1 Feb
  4. 2025RLFactory: A Plug-and-Play Reinforcement Learning Post-Training Framework for LLM Multi-Turn Tool-Use31 Aug

Other references · 15

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  3. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  4. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,182
  5. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIscited by 1,947
  6. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,386
  7. 2020Gradient Surgery for Multi-Task Learningcited by 1,256
  8. 2023GAIA: a benchmark for General AI Assistantscited by 1,128
  9. 2024ToolACE: Winning the Points of LLM Function Callingcited by 216
  10. 2025MCP-Universe: Benchmarking Large Language Models with Real-World Model Context Protocol Serverscited by 96
  11. 2024Hammer: Robust Function-Calling for On-Device Language Models via Function Maskingcited by 55
  12. 2026MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Serverscited by 53
3 more
  1. 2025LiveMCPBench: Can Agents Navigate an Ocean of MCP Tools?cited by 48
  2. 2025MCP-Zero: Active Tool Discovery for Autonomous LLM Agentscited by 33
  3. 2025MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Toolscited by 11

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026A Survey of Agentic Reasoning for Large Language Models: Towards Recursively Self-Improving and Collective Agents
  2. 2026AgentFugue: Agent Scaling for Long-Horizon Tasks through Collective Reasoning
  3. 2025MCP-AgentBench: Evaluating Real-World Language Agent Performance with MCP-Mediated Tools
  4. 2026Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
  5. 2026ProAct: Agentic Lookahead in Interactive Environments
  6. 2025WebCoT: Enhancing Web Agent Reasoning by Reconstructing Chain-of-Thought in Reflection, Branching, and Rollback
  7. 2026MARTI-MARS2: Scaling Multi-Agent Self-Search via Reinforcement Learning for Code Generation
  8. 2025Thinking vs. Doing: Agents that Reason by Scaling Test-Time Interaction