CodeScout: An Effective Recipe for Reinforcement Learning of Code Search Agents

• URL: http://arxiv.org/abs/2603.17829v1
• Date: Wed, 18 Mar 2026 15:25:42 GMT
• Title: CodeScout: An Effective Recipe for Reinforcement Learning of Code Search Agents
• Abstract summary: We show that a coding agent equipped with nothing more than a standard Unix terminal can be trained to achieve strong results.<n>Our work particularly focuses on techniques for re-purposing existing coding agent environments for code search, reward design, and RL optimization.<n>We release the resulting model family, CodeScout, along with all our code and data for the community to build upon.
• Score: 43.426809750160665
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: A prerequisite for coding agents to perform tasks on large repositories is code localization - the identification of relevant files, classes, and functions to work on. While repository-level code localization has been performed using embedding-based retrieval approaches such as vector search, recent work has focused on developing agents to localize relevant code either as a standalone precursor to or interleaved with performing actual work. Most prior methods on agentic code search equip the agent with complex, specialized tools, such as repository graphs derived from static analysis. In this paper, we demonstrate that, with an effective reinforcement learning recipe, a coding agent equipped with nothing more than a standard Unix terminal can be trained to achieve strong results. Our experiments on three benchmarks (SWE-Bench Verified, Pro, and Lite) reveal that our models consistently achieve superior or competitive performance over 2-18x larger base and post-trained LLMs and sometimes approach performance provided by closed models like Claude Sonnet, even when using specialized scaffolds. Our work particularly focuses on techniques for re-purposing existing coding agent environments for code search, reward design, and RL optimization. We release the resulting model family, CodeScout, along with all our code and data for the community to build upon.

Related papers

• PRAXIS: Graph-Grounded Tacit Knowledge for Domain Code Generation [61.752289169878566]
  We identify the root cause of the agent's lack of tacit knowledge, including domain-specific business rules, interface contracts, and operational conventions that developers internalize through practice but never document.<n>This knowledge is deeply buried beneath the domain code, across code entities and their dependency relations, and invisible to the agent that lacks it.<n>We propose PRAXIS, a framework that enables agents to extract, represent, and reuse tacit knowledge for domain code generation.
  arXiv  Detail & Related papers  (2026-08-20T08:28:28Z)
• ToFu: A White-Box, Token-Efficient Agent Harness for Researchers [73.72929185946842]
  ToFu is an agentic harness for researchers that reads your files, edits files, runs commands, and integrates with your development tools.<n>As a research assistant, it supports practical research with superior token efficiency, lower cost, and multilingual capability.<n>As a research object, ToFu provides a white-box agentic harness that allows researchers to inspect, modify, and evaluate its orchestration logic, tool-use behavior, and harness design.
  arXiv  Detail & Related papers  (2026-07-13T11:26:06Z)
• From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intelligence [150.3696990310269]
  Large language models (LLMs) have transformed automated software development by enabling direct translation of natural language descriptions into functional code.<n>We provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs.<n>We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder)
  arXiv  Detail & Related papers  (2025-11-23T17:09:34Z)
• Towards A Generalist Code Embedding Model Based On Massive Data Synthesis [35.04242699869519]
  We introduce textbfCodeR (underlineCode underlineRetrieval), a state-of-the-art embedding model for general-purpose code retrieval.<n>The superior performance of CodeR is built upon CodeR-Pile, a large-scale synthetic dataset constructed under the DRU principle.
  arXiv  Detail & Related papers  (2025-05-19T04:37:53Z)
• Paper2Code: Automating Code Generation from Scientific Papers in Machine Learning [70.04746094652653]
  We introduce PaperCoder, a framework that transforms machine learning papers into functional code repositories.<n>PaperCoder operates in three stages: planning, designs the system architecture with diagrams, identifies file dependencies, and generates configuration files.<n>We then evaluate PaperCoder on generating code implementations from machine learning papers based on both model-based and human evaluations.
  arXiv  Detail & Related papers  (2025-04-24T01:57:01Z)
• Codev-Bench: How Do LLMs Understand Developer-Centric Code Completion? [60.84912551069379]
  We present the Code-Development Benchmark (Codev-Bench), a fine-grained, real-world, repository-level, and developer-centric evaluation framework. Codev-Agent is an agent-based system that automates repository crawling, constructs execution environments, extracts dynamic calling chains from existing unit tests, and generates new test samples to avoid data leakage.
  arXiv  Detail & Related papers  (2024-10-02T09:11:10Z)
• On the Impacts of Contexts on Repository-Level Code Generation [5.641402231731082]
  We present RepoExec, a novel benchmark designed to evaluate repository-level code generation.<n>We focus on three key aspects: executability, functional correctness through comprehensive test case generation, and accurate utilization of cross-file contexts.
  arXiv  Detail & Related papers  (2024-06-17T10:45:22Z)
• CodeAgent: Enhancing Code Generation with Tool-Integrated Agent Systems for Real-World Repo-level Coding Challenges [41.038584732889895]
  Large Language Models (LLMs) have shown promise in automated code generation but typically excel only in simpler tasks. Our research pivots towards evaluating LLMs in a more realistic setting -- real-world repo-level code generation. We present CodeAgent, a novel LLM-based agent framework that employs external tools for effective repo-level code generation.
  arXiv  Detail & Related papers  (2024-01-14T18:12:03Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.