Confucius Code Agent: Scalable Agent Scaffolding for Real-World Codebases

TL;DR

The paper introduces the Confucius Code Agent, a scalable and extensible software engineering agent capable of handling large codebases and complex tasks, with a novel meta-agent for rapid development and refinement.

Contribution

It presents the Confucius SDK and a meta-agent framework that enable scalable, interpretable, and adaptable coding agents for real-world software engineering tasks.

Findings

CCA achieves 59% Resolve@1 on SWE-Bench-Pro, surpassing prior baselines.

The SDK supports long-context reasoning and cross-session learning.

Meta-agent automates agent development and improvement.

Abstract

Real-world software engineering tasks require coding agents that can operate on massive repositories, sustain long-horizon sessions, and reliably coordinate complex toolchains at test time. Existing research-grade coding agents offer transparency but struggle when scaled to heavier, production-level workloads, while production-grade systems achieve strong practical performance but provide limited extensibility, interpretability, and controllability. We introduce the Confucius Code Agent (CCA), a software engineering agent that can operate at large-scale codebases. CCA is built on top of the Confucius SDK, an agent development platform structured around three complementary perspectives: Agent Experience (AX), User Experience (UX), and Developer Experience (DX). The SDK supports a unified orchestrator with advanced context management for long-context reasoning, a persistent note-taking system for cross-session continual learning, and a modular extension system for reliable tool use. In addition, we introduce a meta-agent that automates the construction, evaluation, and refinement of agents through a build-test-improve cycle, enabling rapid agent development on new tasks and tool stacks. Instantiated on the Confucius SDK using the meta-agent, CCA demonstrates strong performance on real-world software engineering tasks. On SWE-Bench-Pro, CCA achieves a Resolve@1 of 59%, exceeding prior research baselines as well as commercial results, under identical repositories, model backends, and tool access.