# Code-QA-Bench: Separating Code Reasoning from Documentation Memorization in Repository-Level QA

arXiv: https://arxiv.org/abs/2605.29277 · HTML: https://mlsanity.com/paper/2605.29277/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 17 of its 33 references that are in the map.
- First posted on arXiv 28 May 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)

## Builds on (most cited first, 17 of 17)

- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code](https://mlsanity.com/paper/2403.07974/) (2024)
- [SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering](https://mlsanity.com/paper/2405.15793/) (2024)
- [A Survey on Large Language Models for Code Generation](https://mlsanity.com/paper/2406.00515/) (2024)
- [CodeSearchNet Challenge: Evaluating the State of Semantic Code Search](https://mlsanity.com/paper/1909.09436/) (2019)
- [From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline](https://mlsanity.com/paper/2406.11939/) (2024)
- [Agentless: Demystifying LLM-based Software Engineering Agents](https://mlsanity.com/paper/2407.01489/) (2024)
- [CRUXEval: A Benchmark for Code Reasoning, Understanding and Execution](https://mlsanity.com/paper/2401.03065/) (2024)
- [SWE-Bench+: Enhanced Coding Benchmark for LLMs](https://mlsanity.com/paper/2410.06992/) (2024)
- [EvoCodeBench: An Evolving Code Generation Benchmark Aligned with Real-World Code Repositories](https://mlsanity.com/paper/2404.00599/) (2024)
- [Prompting Large Language Models to Tackle the Full Software Development Lifecycle: A Case Study](https://mlsanity.com/paper/2403.08604/) (2024)
- [SWE-QA: Can Language Models Answer Repository-level Code Questions?](https://mlsanity.com/paper/2509.14635/) (2025)
- [CodeMind: Evaluating Large Language Models for Code Reasoning](https://mlsanity.com/paper/2402.09664/) (2024)
- [AutoCodeBench: Large Language Models are Automatic Code Benchmark Generators](https://mlsanity.com/paper/2508.09101/) (2025)
- [SWE-QA-Pro: A Representative Benchmark and Scalable Training Recipe for Repository-Level Code Understanding](https://mlsanity.com/paper/2603.16124/) (2026)
- [From Laboratory to Real-World Applications: Benchmarking Agentic Code Reasoning at the Repository Level](https://mlsanity.com/paper/2601.03731/) (2026)

## Built on by (most cited first, 2 of 2)

- [Deep Agentic Search for Repository-Level Code Question Answering: An Empirical Study](https://mlsanity.com/paper/2608.01507/) (2026)
- [Can LLMs Reason About Runtime Behavior? A Repository-Level Dynamic Benchmark](https://mlsanity.com/paper/2609.28449/) (2026)

## Abstract (arXiv)

We present Code-QA-Bench, a fully automated framework for synthesizing repository-level code understanding benchmarks that separates genuine code comprehension from documentation recall and pretraining memorization. The framework makes two methodological contributions: (1) an answer-first generation pipeline where a tool-equipped agent explores source code to produce verified gold answers before deriving questions, ensuring every task is grounded in real code structure; and (2) a three-condition experimental design evaluating agents under closed-book (no repository), code-only (documentation removed), and documented (full repository) conditions, with deltas directly quantifying documentation utility and memorization. We generate 528 code-derivable and 100 doc-dependent tasks across 10 Python repositories from SWE-Bench, scored by an LLM judge on accuracy, completeness, and specificity. Experiments on four frontier models reveal that code access is the dominant factor (+0.23 mean gain over closed-book), documentation provides modest additional benefit (+0.071 on doc-dependent tasks), and code-only ≈ documented on code-derivable tasks, validating the design. The framework is open-source and applicable to any well-documented Python repository.