Appendix B 固定提交哈希
Table 15 列出了每个被分析智能体的仓库 URL 和固定的提交哈希。Section 4 中引用的所有文件路径和行号均指向这些特定提交。读者可以克隆每个仓库并切换到所列提交,以复现或验证本文中的任何论断。
Table 15: 全部 13 个被分析智能体的固定提交哈希。提交按与 Table 1 相同的顺序排列。
| 智能体 | 提交哈希 | 仓库 URL |
|---|---|---|
| Gemini CLI | dd8d4c98b3 | https://github.com/google-gemini/gemini-cli |
| OpenHands | 922e3a2431 | https://github.com/OpenHands/OpenHands |
| Aider | 861a1e4d15 | https://github.com/Aider-AI/aider |
| Cline | 71e312e92a | https://github.com/cline/cline |
| SWE-agent | e72a7e4660 | https://github.com/SWE-agent/SWE-agent |
| Codex CLI | 9dba7337f2 | https://github.com/openai/codex |
| OpenCode | f54abe58cf | https://github.com/anomalyco/opencode |
| Agentless | 5ce5888b9f | https://github.com/OpenAutoCoder/Agentless |
| AutoCodeRover | 585d3e639a | https://github.com/AutoCodeRoverSG/auto-code-rover |
| Moatless Tools | 011ead57a5 | https://github.com/aorwall/moatless-tools |
| Prometheus | b1c722be02 | https://github.com/EuniAI/Prometheus |
| DARS-Agent | eab35168a9 | https://github.com/vaibhavagg303/DARS-Agent |
| mini-swe-agent | 6f1b196616 | https://github.com/SWE-agent/mini-swe-agent |
[^1]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/On/README.md the accuracy of spectrum-based fault localization. In Testing: Academic and Industrial Conference Practice and Research Techniques (TAIC PART), pp. 89–98. Cited by: §4.2.4.
[^2]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/DARS:/README.md dynamic action re-sampling to enhance coding agent performance by adaptive tree traversal. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (ACL), Cited by: §2.3, Table 1.
[^3]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/OpenCode. Note: https://github.com/anomalyco/opencode Cited by: Table 1.
[^4]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Model/README.md Context Protocol. Note: https://modelcontextprotocol.io/ Open specification for connecting AI assistants to external tools and data sources Cited by: §5.6.
[^5]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Claude/README.md Code. Note: https://github.com/anthropics/claude-code The open-source repository contains plugins, examples, and documentation, but the core agent source is distributed as compiled TypeScript bundles in the npm package. Cited by: item 2.
[^6]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/MASAI:/README.md modular architecture for software-engineering AI agents. arXiv preprint arXiv:2406.11638. Cited by: §2.3, item 2.
[^7]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/What's/README.md in a GitHub star? understanding repository starring practices in a social coding platform. Journal of Systems and Software 146, pp. 112–129. Cited by: §1, §3.1, Table 1.
[^8]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/RepairAgent:/README.md an autonomous, LLM-based agent for program repair. In Proceedings of the 47th IEEE/ACM International Conference on Software Engineering (ICSE), pp. 2188–2200. Cited by: §2.2.
[^9]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Understanding/README.md software engineering agents: a study of thought-action-result trajectories. In Proceedings of the 40th IEEE/ACM International Conference on Automated Software Engineering (ASE), Note: arXiv preprint arXiv:2506.18824 Cited by: §2.2, §2.2.
[^10]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/The/README.md anatomy of a large-scale hypertextual web search engine. Computer Networks and ISDN Systems 30 (1–7), pp. 107–117. Cited by: §4.2.4.
[^11]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/A/README.md survey of Monte Carlo tree search methods. IEEE Transactions on Computational Intelligence and AI in Games 4 (1), pp. 1–43. Cited by: §4.1.1.
[^12]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Tree-sitter:/README.md an incremental parsing system for programming tools. Note: https://tree-sitter.github.io/tree-sitter/ Cited by: §4.2.4.
[^13]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Building/README.md AI coding agents for the terminal: scaffolding, harness, context engineering, and lessons learned. arXiv preprint arXiv:2603.05344. Cited by: §1, §1, §2.3, §3.2.
[^14]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Understanding/README.md software engineering agents through the lens of traceability: an empirical study. arXiv preprint arXiv:2506.08311. Cited by: §1, §2.2.
[^15]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Rethinking/README.md the value of agent-generated tests for LLM-based software engineering agents. arXiv preprint arXiv:2602.07900. Cited by: §2.4.
[^16]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/CrewAI. Note: https://github.com/crewAIInc/crewAI Cited by: item 1.
[^17]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/SWE-Bench/README.md Pro: can AI agents solve long-horizon software engineering tasks?. arXiv preprint arXiv:2509.16941. Cited by: §2.4.
[^18]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/SWE-Effi:/README.md re-evaluating software AI agent system effectiveness under resource constraints. arXiv preprint arXiv:2509.09853. Cited by: §2.2, §2.4, §5.2.
[^19]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Bubblewrap:/README.md unprivileged sandboxing tool. Note: https://github.com/containers/bubblewrap Cited by: §4.2.5.
[^20]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/External/README.md Links: Link Cited by: §4.3.1.
[^21]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Configuring/README.md agentic AI coding tools: an exploratory study. arXiv preprint arXiv:2602.14690. Cited by: §2.3, §4.3.4.
[^22]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Saving/README.md SWE-Bench: a benchmark mutation approach for realistic agent evaluation. In Proceedings of the 5th IEEE/ACM International Conference on AI Engineering – Software Engineering for AI (CAIN), Note: arXiv preprint arXiv:2510.08996 Cited by: §2.4, §3.5.
[^23]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Aider:/README.md AI pair programming in your terminal. Note: https://aider.chat/ Cited by: §1, §2.3, Table 1.
[^24]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Gemma:/README.md open models based on Gemini research and technology. arXiv preprint arXiv:2403.08295. Cited by: §4.3.3.
[^25]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Gemini/README.md CLI. Note: https://github.com/google-gemini/gemini-cli Cited by: Table 1.
[^26]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/MetaGPT:/README.md meta programming for a multi-agent collaborative framework. In International Conference on Learning Representations (ICLR), Cited by: item 1.
[^27]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/SWE-bench:/README.md can language models resolve real-world GitHub issues?. In International Conference on Learning Representations (ICLR), Cited by: §1, §2.4, §6.3.
[^28]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Billion-scale/README.md similarity search with GPUs. IEEE Transactions on Big Data 7 (3), pp. 535–547. Cited by: Table 8.
[^29]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Deep/README.md Agents. Note: https://github.com/langchain-ai/deepagents Cited by: item 1.
[^30]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/LangGraph. Note: https://github.com/langchain-ai/langgraph Cited by: Table 2.
[^31]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/The/README.md BrowserGym ecosystem for web agent research. arXiv preprint arXiv:2412.05467. Cited by: §4.2.1.
[^32]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/PEP/README.md 544 – protocols: structural subtyping (static duck typing). Note: https://peps.python.org/pep-0544/ Cited by: §4.4.4.
[^33]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/LlamaIndex. Note: https://github.com/run-llama/llama/_index Cited by: Table 8.
[^34]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Understanding/README.md code agent behaviour: an empirical study of success and failure trajectories. In Proceedings of the 48th IEEE/ACM International Conference on Software Engineering (ICSE), Note: arXiv preprint arXiv:2511.00197 Cited by: §1, §1, §2.2, §2.2, §5.3.
[^35]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/The/README.md landscape of emerging AI agent architectures for reasoning, planning, and tool calling: a survey. arXiv preprint arXiv:2404.11584. Cited by: §1, §2.1, §3.2, §5.1.
[^36]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Neo4j/README.md graph database. Note: https://neo4j.com/ Cited by: §4.2.4.
[^37]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Architectures/README.md for building agentic AI. arXiv preprint arXiv:2512.09458. Cited by: §1, §2.1, §5.1.
[^38]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Open/README.md Interpreter. Note: https://github.com/OpenInterpreter/open-interpreter Cited by: item 1.
[^39]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Codex/README.md CLI. Note: https://github.com/openai/codex Cited by: Table 1.
[^40]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Moatless/README.md Tools. Note: https://github.com/aorwall/moatless-tools Cited by: §1, Table 1.
[^41]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Prometheus:/README.md towards long-horizon codebase navigation for repository-level problem solving. arXiv preprint arXiv:2507.19942. Cited by: §2.2, Table 1.
[^42]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Gorilla:/README.md large language model connected with massive APIs. In Advances in Neural Information Processing Systems (NeurIPS), Cited by: §2.1.
[^43]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Cline. Note: https://github.com/cline/cline Cited by: Table 1, §4.4.5.
[^44]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Guidelines/README.md for conducting and reporting case study research in software engineering. Empirical Software Engineering 14 (2), pp. 131–164. Cited by: §3.4, §6.1, §6.
[^45]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Landlock/README.md LSM: toward unprivileged sandboxing. In Linux Security Summit, Note: https://landlock.io/ Cited by: §4.2.5.
[^46]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/The/README.md coding manual for qualitative researchers. 4th edition, SAGE Publications. Cited by: §3.2, §3.2.
[^47]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Toolformer:/README.md language models can teach themselves to use tools. In Advances in Neural Information Processing Systems (NeurIPS), Cited by: §2.1.
[^48]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Reflexion:/README.md language agents with verbal reinforcement learning. In Advances in Neural Information Processing Systems (NeurIPS), Cited by: §2.1, §5.4.
[^49]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Mastering/README.md the game of Go with deep neural networks and tree search. Nature 529 (7587), pp. 484–489. Cited by: §4.1.1.
[^50]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Toward/README.md architecture-aware evaluation metrics for LLM agents. In Proceedings of the 5th IEEE/ACM International Conference on AI Engineering – Software Engineering for AI (CAIN), Note: arXiv preprint arXiv:2601.19583 Cited by: §1, §2.4, §5.3, §7.
[^51]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Basics/README.md of qualitative research: techniques and procedures for developing grounded theory. 2nd edition, SAGE Publications. Cited by: §3.2, §6.1.
[^52]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/OpenHands:/README.md an open platform for AI software developers as generalist agents. In International Conference on Learning Representations (ICLR), Note: arXiv preprint arXiv:2407.16741 Cited by: §1, §2.3, Table 1.
[^53]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/A/README.md survey on software fault localization. IEEE Transactions on Software Engineering 42 (8), pp. 707–740. Cited by: §4.2.4.
[^54]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Demystifying/README.md LLM-based software engineering agents. Proceedings of the ACM on Software Engineering 2, pp. 801–824. Note: FSE 2025. ACM SIGSOFT Distinguished Paper Award Cited by: §1, §2.3, Table 1.
[^55]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/SWE-Compass:/README.md towards unified evaluation of agentic coding abilities for large language models. arXiv preprint arXiv:2511.05459. Cited by: §2.4, §6.3.
[^56]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/mini-swe-agent. Note: https://github.com/SWE-agent/mini-swe-agent Cited by: §3.1, Table 1.
[^57]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/SWE-agent:/README.md agent-computer interfaces enable automated software engineering. In Advances in Neural Information Processing Systems (NeurIPS), Cited by: §1, §2.3, §3.1, Table 1, §4.2.2, §5.4.
[^58]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/ReAct:/README.md synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR), Cited by: §2.1.
[^59]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/Case/README.md study research and applications: design and methods. 6th edition, SAGE Publications. Cited by: §6.3.
[^60]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/A/README.md survey of learning-based automated program repair. ACM Transactions on Software Engineering and Methodology 33 (2), pp. 1–69. Cited by: §4.2.4.
[^61]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/A/README.md survey on large language models for software engineering. Science China Information Sciences 69, pp. 141102. External Links: Document Cited by: §2.1, §3.1.
[^62]: https://github.com/deusyu/harness-engineering/blob/858c0da6570aad32947c09c7e83fb04f46d22ebe/works/AutoCodeRover:/README.md autonomous program improvement. In Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA), Cited by: §2.3, Table 1.