# ForkPilot: Self-Evolving Policy for Retrospective Search in Long-Horizon Agents

arXiv: https://arxiv.org/abs/2610.04889 · HTML: https://mlsanity.com/paper/2610.04889/

## 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 27 of its 37 references that are in the map.
- First posted on arXiv 4 Oct 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)
- Synthetic data and self-improvement for LLMs / [Self-evolving LLM agents](https://mlsanity.com/synth/self-evolving-llm-agents/)

## Builds on (most cited first, 27 of 27)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [WebArena: A Realistic Web Environment for Building Autonomous Agents](https://mlsanity.com/paper/2307.13854/) (2023)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [AgentBench: Evaluating LLMs as Agents](https://mlsanity.com/paper/2308.03688/) (2023)
- [CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing](https://mlsanity.com/paper/2305.11738/) (2023)
- [Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models](https://mlsanity.com/paper/2310.04406/) (2023)
- [GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning](https://mlsanity.com/paper/2507.19457/) (2025)
- [Group-in-Group Policy Optimization for LLM Agent Training](https://mlsanity.com/paper/2505.10978/) (2025)
- [Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models](https://mlsanity.com/paper/2510.04618/) (2025)
- [Adaptive Conformal Inference Under Distribution Shift](https://mlsanity.com/paper/2106.00170/) (2021)
- [ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory](https://mlsanity.com/paper/2509.25140/) (2025)
- [Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills](https://mlsanity.com/paper/2603.25158/) (2026)
- [An Illusion of Progress? Assessing the Current State of Web Agents](https://mlsanity.com/paper/2504.01382/) (2025)
- [Tree Search for Language Model Agents](https://mlsanity.com/paper/2407.01476/) (2024)
- [WebCanvas: Benchmarking Web Agents in Online Environments](https://mlsanity.com/paper/2406.12373/) (2024)
- [Learning How Hard to Think: Input-Adaptive Allocation of LM Computation](https://mlsanity.com/paper/2410.04707/) (2024)
- [Hindsight Credit Assignment for Long-Horizon LLM Agents](https://mlsanity.com/paper/2603.08754/) (2026)
- [ExACT: Teaching AI Agents to Explore with Reflective-MCTS and Exploratory Learning](https://mlsanity.com/paper/2410.02052/) (2024)
- [VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications](https://mlsanity.com/paper/2509.26490/) (2025)
- [Plan and Budget: Effective and Efficient Test-Time Scaling on Reasoning Large Language Models](https://mlsanity.com/paper/2505.16122/) (2025)
- [Learning When to Plan: Efficiently Allocating Test-Time Compute for LLM Agents](https://mlsanity.com/paper/2509.03581/) (2025)
- [Fact-Augmented Lookahead Planning for LLM Agents](https://mlsanity.com/paper/2506.09171/) (2025)
- [TSR: Trajectory-Search Rollouts for Multi-Turn RL of LLM Agents](https://mlsanity.com/paper/2602.11767/) (2026)
- [Rewarding Beliefs, Not Actions: Consistency-Guided Credit Assignment for Long-Horizon Agents](https://mlsanity.com/paper/2605.20061/) (2026)
- [MM-ToolSandBox: A Unified Framework for Evaluating Visual Tool-Calling Agents](https://mlsanity.com/paper/2607.11818/) (2026)

## Abstract (arXiv)

Interactive language-model agents increasingly solve complex tasks through long-horizon, multi-call reasoning, where errors in beliefs or actions can compound across tool interactions. Retrospective search can recover from such failures but is prone to misallocation. Delayed outcomes obscure the contribution of intermediate search decisions, leading to Attribution Complexity, while evolving execution evidence leads to Adaptation Complexity, where previously learned estimates become stale. To address these challenges, we first introduce Search Value Dynamics (SVD), which characterizes the evolving trade-off between the gain and cost of retrospective search. Building on SVD, we propose ForkPilot, a self-evolving two-stage policy-learning framework. In the first stage, ForkPilot learns a search-value policy offline from completed trajectories through automatically constructed outcome comparisons. In the second stage, it makes search decisions based on current observations and then self-evolves by incorporating newly completed trajectories into subsequent policy updates. We evaluate ForkPilot across 6 diverse benchmarks and 7 widely used LLM backbone families, including four open-source families, GPT-5.6 Sol, and Opus 4.8 in a production agentic system, against 9 competitive baselines, including a real-world harness deployment used by hundreds of thousands of paid users. ForkPilot achieves comparable state-of-the-art performance while reducing token usage by up to 59.2%, demonstrating its efficacy.