# Humanize MLE Agents: Multi-Agent Session Alternation Beyond Single-Agent Limits

arXiv: https://arxiv.org/abs/2610.13494 · HTML: https://mlsanity.com/paper/2610.13494/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use.
- Builds on 17 of its 35 references that are in the map.
- First posted on arXiv 9 Oct 2026.

## Placed in

- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)

## Builds on (most cited first, 17 of 17)

- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation](https://mlsanity.com/paper/2308.08155/) (2023)
- [ChatDev: Communicative Agents for Software Development](https://mlsanity.com/paper/2307.07924/) (2023)
- [ExpeL: LLM Agents Are Experiential Learners](https://mlsanity.com/paper/2308.10144/) (2023)
- [Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models](https://mlsanity.com/paper/2310.04406/) (2023)
- [The Option-Critic Architecture](https://mlsanity.com/paper/1609.05140/) (2016)
- [AFlow: Automating Agentic Workflow Generation](https://mlsanity.com/paper/2410.10762/) (2024)
- [MLEvolve: A Self-Evolving Framework for Automated Machine Learning Algorithm Discovery](https://mlsanity.com/paper/2606.06473/) (2026)
- [JIT-Agent: Scaling Harness Intelligence via Just-in-Time Harness Evolution](https://mlsanity.com/paper/2608.25593/) (2026)
- [AgentGit: A Version Control Framework for Reliable and Scalable LLM-Powered Multi-Agent Systems](https://mlsanity.com/paper/2511.00628/) (2025)
- [Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesses](https://mlsanity.com/paper/2608.24876/) (2026)
- [Learning with Options that Terminate Off-Policy](https://mlsanity.com/paper/1711.03817/) (2017)
- [The Handoff Tax: Continuing Non-Native Trajectories in LLM Agents](https://mlsanity.com/paper/2608.24358/) (2026)
- [CAKE: Compiler-Agent Co-Design for Frontier Kernel Evolution](https://mlsanity.com/paper/2608.12629/) (2026)
- [Relay, Don't Route: Adaptive Population Handoff for Cost-Efficient LLM-Driven Evolution](https://mlsanity.com/paper/2608.05651/) (2026)
- [Learning to Hand Off: Provably Convergent Workflow Learning under Interface Constraints](https://mlsanity.com/paper/2605.19140/) (2026)
- [Humanize: Judgement Engineering for Agentic Coding](https://mlsanity.com/paper/2610.08900/) (2026)

## Abstract (arXiv)

Existing LLM-based agent approaches for long-horizon machine learning engineering (MLE) often use context engineering, implemented through custom harnesses that manage tools, skills, and accumulated experience. However, we find that as single-session execution lengthens, the agent's reasoning intensity declines alongside slowing progress, summarized as long-horizon deceleration, and the benefits of these custom harnesses vanish as the underlying model improves, summarized as harness staleness. These findings suggest that their benefits may not persist across versions, which motivates a workflow-level approach in which agents build on one another's work across sessions. We model these alternating agent sessions as a context-reset semi-Markov decision process (CR-SMDP) and establish sufficient conditions for alternation to exceed single-agent reference limits, derive an explicit experiment-budget bound, and characterize when session caps preserve this advantage. Guided by this analysis, we introduce Humanize MLE Agents (HMA), which alternates two agents on a shared workspace, limits each session to at most k experiments and resets the context between sessions, without additional task-specific harness or prior knowledge supplements. Across 75 Kaggle tasks in MLE-bench, HMA reaches a state-of-the-art any-medal rate of 78.2%, exceeding the strongest open-source baseline by 8.0 points with a quarter of its time budget. Moreover, among six configurations spanning five heterogeneous agent pairs, HMA improves any-medal rate over the constituent baselines by 6.4 points on average, demonstrating cross-pair generality. A supplementary case on kernel optimization reaches 878 cycles, within 1.6% of the community leaderboard leader, demonstrating applicability beyond MLE.