Results suggest ContextSniper can substantially cut token and cost overhead for repository-level repair agents without a measurable loss in repair quality, part of AntTrail's broader agent-memory engine.
Abstract
Large language model agents can repair real repository issues, but they often spend large context budgets on whole-file reads, broad searches, and long terminal outputs where useful evidence is mixed with irrelevant code and logs. This paper presents ContextSniper, AntTrail's code-repair module for precision evidence selection in repository-level program repair, part of AntTrail's broader agent-memory engine. AntTrail is available at https://gitcode.com/datagallery/AntTrail. ContextSniper indexes code and action memory as three abstract levels, retrieves candidates with a hybrid ranker, filters long tool output through an intention-aware context gate, and returns compact evidence packets while keeping full source recoverable on demand. In a matched 50-task-per-condition comparison on SWE-bench Lite (same tasks, baseline vs.\ ContextSniper), ContextSniper reduces total token use by 51.5% and logged cost by 36.4% for OpenClaw, and by 38.9% and 27.3% for Claude Code, with submitted-resolution rates essentially unchanged in both host-agent settings. In a separate five-task comparison, ContextSniper beats existing memory- and RAG-style integrations on token efficiency. These results suggest ContextSniper can substantially cut token and cost overhead for repository-level repair agents without a measurable loss in repair quality. The evaluation harness for this study is available at https://gitcode.com/lukchiwang/ContextSniper.
The ORCA-bench benchmark is introduced, a benchmark that puts general-purpose coding agents in a production-fidelity oncall setting and is a lower bound on the engineering investment required before frontier coding agents can be safely entrusted with production reliability.
Albert Gong, Kyuseong Choi, Abhineet Agarwal et al.· 0 citations
Unit testing is crucial for software reliability in JavaScript, yet constructing comprehensive suites remains costly and error-prone. JavaScript’s dynamic features, flexible typing, asynchronous execution, and reliance on third-party libraries, pose persistent challenges for automated test generation. Although large language models (LLMs) show promise in code reasoning and synthesis, existing methods often miss the multilayer contextual information required for executable, semantically correct tests. This paper introduces JSTestCraft, an agentic, multi-layer framework for adaptive JavaScript unit test generation. JSTestCraft reconstructs missing context via three enrichment agents: library, structural, and semantic. These agents capture inter-function topology, third-party API semantics, and inferred type constraints in a shared contextual memory for reasoning-driven test synthesis. A testing and optimization layer generates, executes, and iteratively refines test cases. Evaluated on 20 real-world Node.js repositories, JSTestCraft outperforms state-of-the-art baselines, improving test pass rate by 60.9%, statement coverage by 14.2%, and branch coverage by 47.3%. Ablation analysis confirms each agent’s contribution to contextual completeness. Beyond metrics, JSTestCraft discovers 13 previously unknown bugs across 5 repositories, 6 of which received substantive acknowledgment from maintainers or community contributors. These results demonstrate that context reconstruction and agentic collaboration enable LLMs to perform more reliable and adaptive testing in dynamic JavaScript environments.
Enterprise agents must follow long-horizon, conditional, safety-critical standard operating procedures (SOPs), which are compiled into executable pseudo-code and run with a program-guided stack machine that pages the active frame while an LLM performs semantic execution.
Chenglin Yu, Lichao Yin, Qingxin Fan et al.· 1 citation
Developers increasingly delegate real maintenance work to product-grade coding agents, and many state tasks in their native language, in the style of a customer request rather than a curated English issue. We introduce RuBench 1.0, a benchmark of 25 tasks mined from recent fix commits in five live open-source repositories (aiohttp, aiogram, Laravel, NestJS, Fastify), each specified natively in Russian -- written from scratch, not translated -- and judged by the upstream maintainer's regression tests, which we withhold from release. All fix commits postdate the training-data cutoffs of every evaluated model. Round 1 evaluates Claude Code with Opus 4.8, Sonnet 5, and Haiku 4.5, and Codex CLI with GPT-5.5 (3 independent runs each; pass@1 with task-level uncertainty); the best configuration resolves 78.7% of tasks. Auditing full trajectories of an hors-concours configuration (Claude Code + Fable 5), we caught the product silently substituting the model on 20% of tasks via an official safeguard fallback -- evidence that the deployed product, not the model, is the unit actually measured. Version 2 adds Round 2: seven further configurations on the same frozen set under a per-configuration freshness gate -- the Russian-market agents SourceCraft CLI (ds, legacy) and Koda CLI (koda-pro), Antigravity with Gemini 3.1 Pro and 3.5 Flash, and Codex CLI with GPT-5.6 Sol and Luna. SourceCraft's flagship resolves 68.1% (N=23), above GPT-5.5 and both Gemini rows. A tool-call contamination re-audit of all 437 Round-2 trajectories finds the Russian and Gemini columns clean (0/293 cells) while flagging systematic oracle-hunting in the GPT-5.6 family (8/69 and 13/75 cells), including one case of mining a prior round's artifacts from the run machine's disk; honest scores are published alongside raw ones. We release statements, metadata, trajectories, and diffs; oracles are withheld with a SHA-256 manifest.
These results establish semantic prompt structure as a robust signal for KV-cache management while clarifying how it should be combined with attention-based importance.
PerfAgent is presented, a profiler-guided, verifier-in-the-loop workflow that gives an off-the-shelf coding agent the feedback needed to find real hotspots, improve beyond the first passing patch, and use profiler evidence rather than timing alone to decide what to optimize next.
Ryan Deng, Yuanzhe Liu, Bastian Lipka et al.· 2 citations
