ProjAgent: Procedural Similarity Retrieval for Repository-Level Code Generation

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: QiHong Chen, Aaron Imani, Iftekhar Ahmed
• Venue: arXiv: cs.SE
• Published: 2026-07-09
• Collected: 2026-07-10
• URL: http://arxiv.org/abs/2607.08691v1
• PDF: https://arxiv.org/pdf/2607.08691v1
• DOI: UNKNOWN

Plain-Language Brief

The paper proposes ProjAgent, a system for generating code at the repository level. It introduces 'procedural similarity' as a new way to retrieve relevant functions, addressing a limitation of existing methods that overlook procedural logic in different identifiers or application domains. ProjAgent performs better than existing retrieval-based baselines on the REPOCOD dataset.

What the Document Says

• ProjAgent is a system for repository-level code generation.
• It introduces procedural similarity as a new retrieval signal.
• ProjAgent decomposes the target function into intermediate steps and retrieves functions with similar procedural behavior.
• It integrates procedural context with conventional semantic retrieval for richer repository context.
• ProjAgent includes a conservative static-analysis feedback loop for code repair.
• The system was evaluated on REPOCOD, achieving 41.14% Pass@1, outperforming existing retrieval-based baselines.

Key Concepts Mentioned

• ProjAgent
• repository-level code generation
• procedural similarity
• intermediate reasoning steps
• agentic workflow
• REPOCOD

Claims Made by the Document

• Existing retrieval methods overlook procedural logic.
• Procedural similarity is an effective and previously unexplored retrieval dimension for repository-level code generation.

Evidence Used

• The paper discusses the limitations of existing retrieval methods and the introduction of ProjAgent as a solution.
• The paper reports the results of ProjAgent on REPOCOD, showing better performance than existing baselines.

Methods / Approach

• Decomposition of target function into intermediate steps
• Employment of an agentic workflow
• Static-analysis feedback loop for code repair

Limitations Stated by the Authors

Not stated in source

Limitations Noted by the Agent

Agent interpretation — not from the source.

The performance of ProjAgent might be affected by the quality and coverage of the procedural logic in the repository.

Open Questions

• How does ProjAgent handle complex, multi-step procedures?
• What is the impact of ProjAgent on code maintainability and readability?

Notes for Future Q&A

May include interpretation. Labeled where it does.

The text focuses on the introduction of ProjAgent and its performance, but it doesn't provide detailed information about how it handles complex procedures or its effects on code quality.

Raw Abstract / Excerpt

Repository-level code generation requires implementing target functions while accounting for complex cross-file dependencies and project-specific conventions. Existing retrieval methods predominantly rely on lexical, structural, or semantic similarity, often overlooking repository functions that implement similar procedural logic despite differing in identifiers or application domains. We propose ProjAgent, a repository-level code generation system that introduces procedural similarity as an explicit retrieval signal. ProjAgent decomposes the target function into intermediate reasoning steps and employs an agentic workflow to retrieve repository functions that exhibit similar procedural behavior at each step. The retrieved procedural context is integrated with conventional semantic retrieval to construct a richer repository context for code generation. ProjAgent further incorporates a conservative static-analysis feedback loop that iteratively repairs generated code using compiler and static-analysis feedback. Evaluated on REPOCOD, ProjAgent achieves 41.14% Pass@1, outperforming existing retrieval-based baselines. These results demonstrate that procedural similarity is an effective and previously unexplored retrieval dimension for repository-level code generation.