CodeSpec: Dual Executable Specifications for Agentic Long-Horizon Feature Development

Source: https://arxiv.org/abs/2607.26777

Authors: Peiding Wang, Li Zhang, Fang Liu, Taichuan Li, Yinghao Zhu

Published: 2026-07-29

Categories: cs.SE

Abstract

LLM-based code agents have advanced repository-level software development through iterative interaction with codebases and tools. However, feature development requires integrating new behaviors into existing architectures through coherent cross-component functional chains. Existing agents typically derive such chains through free-form reasoning, often producing unreliable feature designs with incomplete functional chains. Moreover, textual designs are difficult to verify and enforce, making it challenging to maintain design-implementation consistency throughout long-horizon development.

CodeSpec is a dual executable specification method for repository-level feature development. It builds reliable functional chains from evidence pairing sub-requirement semantics with repository architectures, then compiles them into complementary architecture and behavior specifications that check chain completeness and correctness while preserving design-implementation consistency over long interactions.

On FeatureBench, CodeSpec achieves 70.7%, 55.0%, and 49.9% pass rates under DeepSeek-V4-Pro, outperforming representative baselines such as Claude Code. Results on NL2Repo-Bench further demonstrate its generalizability.

Evidence (Obsidian)

From 论文流水线 2026-07-31:

双规格分工：architecture specification 检查链的完整性，behavior specification 检查正确性。
和 SpecFirst（spec 前置 elicitation）可对照读：一个用双规格贯穿长程开发，一个把 spec elicitation 独立成前置阶段。