# Act More, Decide Less: Skill-Guided Adaptive Action Chunking for Long-Horizon LLM Agents

arXiv: https://arxiv.org/abs/2609.02042 · HTML: https://mlsanity.com/paper/2609.02042/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 28 of its 36 references that are in the map.
- First posted on arXiv 2 Sep 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 28 of 28)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [ALFWorld: Aligning Text and Embodied Environments for Interactive Learning](https://mlsanity.com/paper/2010.03768/) (2020)
- [Group-in-Group Policy Optimization for LLM Agent Training](https://mlsanity.com/paper/2505.10978/) (2025)
- [UniVLA: Learning to Act Anywhere with Task-centric Latent Actions](https://mlsanity.com/paper/2505.06111/) (2025)
- [TravelPlanner: A Benchmark for Real-World Planning with Language Agents](https://mlsanity.com/paper/2402.01622/) (2024)
- [ScienceWorld: Is your Agent Smarter than a 5th Grader?](https://mlsanity.com/paper/2203.07540/) (2022)
- [SkillRL: Evolving Agents via Recursive Skill-Augmented Reinforcement Learning](https://mlsanity.com/paper/2602.08234/) (2026)
- [ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL](https://mlsanity.com/paper/2402.19446/) (2024)
- [Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills](https://mlsanity.com/paper/2603.25158/) (2026)
- [Self-Imitation Learning](https://mlsanity.com/paper/1806.05635/) (2018)
- [Where LLM Agents Fail and How They can Learn From Failures](https://mlsanity.com/paper/2509.25370/) (2025)
- [Inducing Programmatic Skills for Agentic Tasks](https://mlsanity.com/paper/2504.06821/) (2025)
- [Reinforcement Learning for Long-Horizon Interactive LLM Agents](https://mlsanity.com/paper/2502.01600/) (2025)
- [SPA-RL: Reinforcing LLM Agents via Stepwise Progress Attribution](https://mlsanity.com/paper/2505.20732/) (2025)
- [AutoManual: Constructing Instruction Manuals by LLM Agents via Interactive Environmental Learning](https://mlsanity.com/paper/2405.16247/) (2024)
- [Meta-RL Induces Exploration in Language Agents](https://mlsanity.com/paper/2512.16848/) (2025)
- [RetroAgent: From Solving to Evolving via Retrospective Dual Intrinsic Feedback](https://mlsanity.com/paper/2603.08561/) (2026)
- [ReCode: Unify Plan and Action for Universal Granularity Control](https://mlsanity.com/paper/2510.23564/) (2025)
- [Policy and World Modeling Co-Training for Language Agents](https://mlsanity.com/paper/2606.02388/) (2026)
- [TimeBlind: A Spatio-Temporal Compositionality Benchmark for Video LLMs](https://mlsanity.com/paper/2602.00288/) (2026)
- [Adapt2Reward: Adapting Video-Language Models to Generalizable Robotic Rewards via Failure Prompts](https://mlsanity.com/paper/2407.14872/) (2024)
- [Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model](https://mlsanity.com/paper/2603.25184/) (2026)
- [SEAR: Sample Efficient Action Chunking Reinforcement Learning](https://mlsanity.com/paper/2603.01891/) (2026)
- [Dataset Condensation for Recommendation](https://mlsanity.com/paper/2310.01038/) (2023)
- [Seeing Farther and Smarter: Value-Guided Multi-Path Reflection for VLM Policy Optimization](https://mlsanity.com/paper/2602.19372/) (2026)
- [DARE: Difficulty-Adaptive Reinforcement Learning with Co-Evolved Difficulty Estimation](https://mlsanity.com/paper/2605.09188/) (2026)
- [WatchAct: A Benchmark for Behavior-Grounded Robot Manipulation](https://mlsanity.com/paper/2606.26443/) (2026)

## Built on by (most cited first, 1 of 1)

- [ChunkAct: Action Chunking for Agentic Robotics](https://mlsanity.com/paper/2610.18874/) (2026)

## Abstract (arXiv)

Large language model (LLM) agents for long-horizon interactive tasks typically follow a ReAct-style protocol, issuing one primitive action per LLM round. While this enables frequent replanning, it is inefficient for long-horizon tasks where many rounds are spent on routine action sequences. A natural alternative is to let the agent emit variable-length action chunks. However, naively training such policies with standard reinforcement learning fails: the agent either collapses to single-action behavior or over-commits to excessively long sequences. Both failures share a common root cause: the inability to learn chunk boundaries. We propose SPACE, which addresses this challenge by distilling chunk-boundary supervision from trajectory-induced programmatic skills. We induce two-level programmatic skills from successful trajectories, where subskill boundaries serve as direct chunk-boundary supervision. This temporal structure is then distilled into a primitive-chunk policy via hybrid on-/off-policy optimization with chunk-aware credit assignment. Experiments on ALFWorld and ScienceWorld show that SPACE improves success rates by 7.0%-31.3% over the strongest baseline in each setting while reducing average LLM decision rounds by up to 78.9%.