5arXiv cs.CL (Computation and Language)·
SWE-Pruner Pro uses agent's internal representations for efficient context pruning in coding agents
SWE-Pruner Pro is a new context pruning method for coding agents that leverages the agent's own internal representations to decide which lines of tool output to keep or discard, rather than relying on a separate classifier. A lightweight head converts these internal representations into keep-or-prune labels per line, augmented with a length-aware embedding. Evaluated across two open-weight backbones and four multi-turn benchmarks, the method saves up to 39% of prompt and completion tokens while maintaining task quality, and on MiMo-V2-Flash additionally improves SWE-Bench Verified resolve rate by +3.8%.
Related guides (3)
Related events (8)
TRIM algorithm reduces AI-generated code bloat by minimizing agent trajectories
Researchers introduce 'CodeSlop' as a formal concept describing residual, functionally unnecessary edits that accumulate in AI-generated code as coding agents iterate toward solutions. The paper presents TRIM (Trajectory-guided Redundancy Identification and Minimization), which reduces CodeSlop by 17.9–32.9% across agentic scaffolds by minimizing agent trajectories rather than targeting redundant code directly. TRIM achieves this with roughly half the validation cost of baseline approaches like Delta Debugging and negligible performance regression, addressing a compounding maintainability problem as agents take on larger shares of codebases.
SelfCompact: Model-driven adaptive context compaction for long agent traces
Researchers propose SelfCompact, a scaffold that lets language models decide when and how to compact their own accumulated context during long agentic runs, rather than relying on fixed token-threshold triggers. The system pairs a compaction tool with a lightweight rubric specifying when to invoke or suppress compaction based on trajectory structure (e.g., sub-task completion vs. mid-derivation). Evaluated across six benchmarks and seven models, SelfCompact matches or exceeds fixed-interval summarization while reducing per-question token cost by 30-70%, with gains of up to 18.1 points on math tasks and 5-9 points on agentic search. The work identifies a 'meta-cognitive gap' in unprompted models and shows it can be closed via scaffolding without fine-tuning.
Persona-Pruner: framework for sculpting lightweight persona-specific LLM sub-networks via structured pruning
Persona-Pruner is a pruning framework that isolates persona-specific sub-networks from a generalist language model given only a character description, producing lightweight role-playing models without the full model's computational cost. The authors observe that naive pruning degrades role-playing fidelity by failing to distinguish redundant knowledge from character-essential parameters. On RoleBench, Persona-Pruner reduces performance drop by up to 93.8% relative to the strongest baseline pruning method while preserving general LLM capabilities. The work targets practical deployment scenarios such as game ecosystems with many simultaneous NPC agents.
SWE-Explore: New benchmark isolates repository exploration capability in coding agents
SWE-Explore is a new benchmark targeting repository exploration as a distinct, fine-grained capability of coding agents, separate from end-to-end task resolution. It covers 848 issues across 10 programming languages and 203 open-source repositories, with line-level ground truth derived from successful agent trajectories. Evaluation across retrieval methods, coding agents, and specialized localizers finds that agentic explorers outperform classical retrieval, and that line-level coverage and efficient ranking remain the key differentiators at the frontier. The benchmark addresses a gap in SWE-bench-style evaluations that treat task resolution as a binary outcome.
SWE-Interact benchmark evaluates coding agents on multi-turn, user-driven software engineering tasks
SWE-Interact is a new benchmark testbed that evaluates coding agents in realistic multi-turn developer workflows, where a user simulator starts with vague instructions and progressively reveals requirements. Unlike existing SWE benchmarks that provide complete specs upfront, SWE-Interact tests interactive goal discovery and iterative refinement. Frontier models including Claude Opus 4.8 and GPT-5.5 solve ~50% of single-turn baseline tasks but only ~25% of SWE-Interact tasks, revealing a significant capability gap. The benchmark is grounded in large-scale studies of real coding-agent interactions and identifies failure modes like over-agentic coding, requirement forgetting, and early abandonment under ambiguity.
Probe-and-Refine Tuning improves coding agent performance via iterative repository guidance refinement
A new arXiv paper introduces probe-and-refine tuning, a procedure that uses synthetic bug-fix probes to iteratively improve AGENTS.md repository guidance files for LLM-based coding agents without requiring an agent loop during tuning. Evaluated on SWE-bench Verified with Qwen3.5-35B-A3B, the method achieves 33.0% mean resolve rate versus 28.3% for a static knowledge base baseline and 25.5% for an unguided baseline. The improvement is attributed to coverage gains—refined guidance helps agents locate the correct files rather than improving patch quality—and a step-budget experiment shows guidance is necessary for agents to productively use larger compute budgets.
DeepSWE, ProgramBench, and ITBench-AA emerge as harder successors to SWE-bench for agent evaluation
Three new benchmarks — DeepSWE (by Datacurve), ProgramBench (Meta/Stanford/Harvard), and ITBench-AA (IBM/Artificial Analysis) — are positioned as more rigorous replacements for the SWE-bench family, which models have largely saturated. DeepSWE tests feature implementation using private codebases and human-written problems; ProgramBench evaluates agents' ability to recreate functional programs from scratch; ITBench-AA measures root-cause diagnosis in real-world IT incident scenarios. Current top performers include GPT-5.5 (70% on DeepSWE), Claude Opus 4.7 (46.7% on ITBench-AA), and Claude Opus 4.7 (3% on ProgramBench at the 95% pass threshold), illustrating that even frontier models have substantial headroom.
ProjAgent introduces procedural similarity retrieval for repository-level code generation
ProjAgent is a new repository-level code generation system that adds procedural similarity as an explicit retrieval signal, decomposing target functions into intermediate reasoning steps to find analogous implementations across a codebase. It combines this with conventional semantic retrieval and a static-analysis feedback loop for iterative code repair. Evaluated on the REPOCOD benchmark, ProjAgent achieves 41.14% Pass@1, outperforming existing retrieval-based baselines. The work identifies procedural similarity as a previously unexplored retrieval dimension for this class of coding agents.