REVIEW 34 cited by
SWE-Search: Enhancing Software Agents with Monte Carlo Tree Search and Iterative Refinement
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Software engineers operating in complex and dynamic environments must continuously adapt to evolving requirements, learn iteratively from experience, and reconsider their approaches based on new insights. However, current large language model (LLM)-based software agents often follow linear, sequential processes that prevent backtracking and exploration of alternative solutions, limiting their ability to rethink their strategies when initial approaches prove ineffective. To address these challenges, we propose SWE-Search, a multi-agent framework that integrates Monte Carlo Tree Search (MCTS) with a self-improvement mechanism to enhance software agents' performance on repository-level software tasks. SWE-Search extends traditional MCTS by incorporating a hybrid value function that leverages LLMs for both numerical value estimation and qualitative evaluation. This enables self-feedback loops where agents iteratively refine their strategies based on both quantitative numerical evaluations and qualitative natural language assessments of pursued trajectories. The framework includes a SWE-Agent for adaptive exploration, a Value Agent for iterative feedback, and a Discriminator Agent that facilitates multi-agent debate for collaborative decision-making. Applied to the SWE-bench benchmark, our approach demonstrates a 23% relative improvement in performance across five models compared to standard open-source agents without MCTS. Our analysis reveals how performance scales with increased inference-time compute through deeper search, providing a pathway to improve software agents without requiring larger models or additional training data. This highlights the potential of self-evaluation driven search techniques in complex software engineering environments.
Forward citations
Cited by 34 Pith papers
-
LLM Agents Can See Code Repositories
Visual graphs of repository structure added to text inputs for multimodal LLM agents reduce token consumption by up to 26% while maintaining or improving issue-resolution accuracy.
-
AgentLens: Revealing The Lucky Pass Problem in SWE-Agent Evaluation
10.7% of passing SWE-agent trajectories are Lucky Passes with chaotic behaviors, and a quality score based on process references changes model rankings across eight backends.
-
AgentLens: Revealing The Lucky Pass Problem in SWE-Agent Evaluation
AgentLens reveals 10.7% of passing SWE-agent trajectories exhibit Lucky Pass behaviors and introduces a process-level evaluation framework with a new annotated dataset of 1,815 trajectories.
-
Can You Trust the Vectors in Your Vector Database? Black-Hole Attack from Embedding Space Defects
Injecting a few malicious vectors near the centroid exploits centrality-driven hubness in high-dimensional embeddings, causing them to dominate top-k retrievals in up to 99.85% of cases.
-
Dynamic analysis enhances issue resolution
DAIRA integrates dynamic tracing into LLM agents to achieve 79.4% resolution rate on SWE-bench Verified for code defect repair.
-
Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
Evo-Memory is a new benchmark for self-evolving memory in LLM agents across task streams, with baseline ExpRAG and proposed ReMem method that integrates reasoning, actions, and memory updates for continual improvement.
-
When Does Restricting a Coding Agent to execute_code Help? A Regime $\times$ Agent-Design Ablation
Code-only tool surfaces beat or match the cheapest tool-rich rival on cost in three of four (regime, agent) cells at matched pass rates; Claude on SWE-bench is the directional exception via failure cost.
-
Autonomous Repair for Multi-Agent Systems via Monte-Carlo Tree Search
An MCTS-based repair loop with partial rollouts and an LLM-as-a-judge reward raises post-repair pass rates by 3–12% over DoVer, ReAct, and Reflexion baselines, backed by a new replayable benchmark StateMAS.
-
PhoenixRepair: Rethinking Repair Strategy Exploration in Software Agents
A multi-agent repair framework that samples multiple edit locations and iteratively reflects on patch attempts reaches 76.0% Pass@1 on SWE-bench-Verified, up to a 7.8% relative gain over SWE-agent.
-
IssueExec: A Test-Driven Approach for Localizing Software Engineering Issues
IssueExec localizes issue-relevant code by retrieving related tests and following their execution traces, beating prior localization methods on SWE-bench Lite.
-
A Single Patch Is Not Enough: Deterministic Fusion of Repair Candidates
PatchFusion uses deterministic atomic evidence fusion on candidate patches to outperform ranking, test-filtering, and LLM-judge selectors on SWE-bench and Defects4J pools.
-
LLM Agents Can See Code Repositories
Adding visual dependency-graph images to a text-based coding agent cuts token consumption by up to 26% while keeping issue-resolution accuracy roughly unchanged.
-
STRIDE: Learnable Stepwise Language Feedback for LLM Reasoning
STRIDE co-trains generator and verifier on outcome rewards alone to deliver learnable stepwise language feedback that redirects LLM reasoning trajectories and outperforms scalar-reward baselines.
-
Step Rejection Fine-Tuning: A Practical Distillation Recipe
Step Rejection Fine-Tuning masks loss on erroneous steps identified by a critic LLM in unresolved trajectories, raising SWE-bench Verified resolution rate by 3.7% to 32.2% versus 2.4% for trajectory-level rejection.
-
HELO-APR: Enhancing Low-Resource Program Repair through Cross-Lingual Knowledge Transfer
HELO-APR improves LLM-based automatic program repair in low-resource languages by synthesizing cross-lingual training data and using curriculum learning, raising Pass@1 from 31.32% to 48.65% on DeepSeek-Coder for Ruby...
-
GALA: Multimodal Graph Alignment for Bug Localization in Automated Program Repair
GALA uses hierarchical graph alignment between UI screenshots and code structures to achieve state-of-the-art bug localization in multimodal automated program repair on SWE-bench.
-
On the Role of Fault Localization Context for LLM-Based Program Repair
More fault localization context does not consistently improve LLM-based program repair; file-level context gives 15-17x gains, optimal around 6-10 files, while line-level context often degrades performance from noise.
-
Can You Trust the Vectors in Your Vector Database? Black-Hole Attack from Embedding Space Defects
Injecting a few vectors near the embedding-space centroid can make them appear in top-k results for up to 94.4% of queries via centrality-driven hubness.
-
Beyond Fixed Tests: Repository-Level Issue Resolution as Coevolution of Code and Behavioral Constraints
Agent-CoEvo is a multi-agent LLM framework that coevolves code patches and test patches to resolve repository-level issues, outperforming fixed-test baselines on SWE-bench Lite and SWT-bench Lite.
-
Dynamic analysis enhances issue resolution
Embedding dynamic analysis into an LLM repair agent yields a claimed 79.4% resolution rate on SWE-bench Verified while cutting token use by about 25%.
-
SGAgent: Suggestion-Guided LLM-Based Multi-Agent Framework for Repository-Level Software Repair
A three-agent locate-suggest-fix framework with a knowledge-graph toolkit resolves 154/300 SWE-Bench-Lite issues with Claude-3.5, outperforming same-model baselines by 5-10 points.
-
ToolSelf: Unifying Task Execution and Self-Reconfiguration via Tool-Driven Emergent Adaptation
An LLM agent that can call a reconfiguration tool to update its sub-goals, toolbox, strategy, and context outperforms static-config agents across FRAMES, xbench, GAIA, and SWE-bench Lite.
-
SVRepair: Structured Visual Reasoning for Automated Program Repair
A multimodal program-repair system that converts bug screenshots into semantic scene graphs and iteratively crops to bug regions, reporting 36.47% on SWE-Bench M, 38.02% on MMCode, and 95.12% on CodeVision.
-
From Historical Patches to Repair Plans: Outcome-Conditioned Reasoning for Repository-Level Program Repair
Backward reasoning distilled from a repository's already-fixed issues, injected as prompts, lifts LLM repair success on SWE-Bench Lite by up to 10.4 Pass@1 points.
-
Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
Evo-Memory is a new streaming benchmark and evaluation framework for self-evolving memory in LLM agents, unifying over ten memory modules and introducing the ReMem pipeline for continual improvement on multi-turn and ...
-
SWE-Debate: Competitive Multi-Agent Debate for Software Issue Resolution
A competitive multi-agent debate over graph-derived fault propagation chains lifts automated issue resolution on SWE-bench Verified to 41.4% pass@1.
-
SWE-Perf: Can Language Models Optimize Code Performance on Real-World Repositories?
A repository-level benchmark shows LLMs and agents currently achieve far smaller, statistically verified speedups than expert human patches on 140 real-world Python performance tasks.
-
Is Inter-Seed Cross-Play Enough? Evaluating the Robustness of Zero-Shot Coordination Algorithms to Implementation Details
For Other-Play in Yokai, agents trained with different implementation details coordinate across implementations about as well as across seeds, supporting inter-seed cross-play as a proxy for cross-implementation evaluation.
-
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application
This survey categorizes agentic environments for LLMs by eight attributes and domains, introduces symbolic and neural synthesis paradigms with evaluation, and outlines four agent evolution pathways plus three environm...
-
Code as Agent Harness
A survey that organizes existing work on LLM-based agents around code as the central harness, structured in three layers of interfaces, mechanisms, and multi-agent scaling, with applications across domains and listed ...
-
HELO-APR: Enhancing Low-Resource Program Repair through Cross-Lingual Knowledge Transfer
HELO-APR uses cross-lingual synthetic data generation and curriculum learning to improve LLM-based automatic program repair in low-resource programming languages.
-
Bias in the Loop: Auditing LLM-as-a-Judge for Software Engineering
LLM judges for code tasks show high sensitivity to prompt biases that systematically favor certain options, changing accuracy and model rankings even when code is unchanged.
-
Detection Time Distribution Predicted Using Absorbing Boundary Conditions and Imaginary Potentials
Three inequivalent quantum detection-time proposals yield waveguide arrival distributions showing partial reflection, spin-orientation independence (for tested parameters), and width dependence when spin couples to th...
-
Agentic Reasoning for Large Language Models
The survey structures agentic reasoning for LLMs into foundational, self-evolving, and collective multi-agent layers while distinguishing in-context orchestration from post-training optimization and reviewing applicat...
Discussion (0). Sign in to comment.