AgentForge: Execution-Grounded Multi-Agent LLM Framework for Autonomous Software Engineering
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866913032423079936 |
|---|---|
| author | Kumar, Rajesh Ali, Waqar Ahmed, Junaid Ali, Najma Imtiaz Usman, Shaban |
| author_facet | Kumar, Rajesh Ali, Waqar Ahmed, Junaid Ali, Najma Imtiaz Usman, Shaban |
| contents | Large language models generate plausible code but cannot verify correctness. Existing multi-agent systems simulate execution or leave verification optional. We introduce execution-grounded verification as a first-class principle: every code change must survive sandboxed execution before propagation. We instantiate this principle in AGENTFORGE, a multi-agent framework where Planner, Coder, Tester, Debugger, and Critic agents coordinate through shared memory and a mandatory Docker sandbox. We formalize software engineering with LLMs as an iterative decision process over repository states, where execution feedback provides a stronger supervision signal than next-token likelihood. AGENTFORGE achieves 40.0\% resolution on SWE-BENCH Lite, outperforming single-agent baselines by 26--28 points. Ablations confirm that execution feedback and role decomposition each independently drive performance. The framework is open-source at https://github.com/raja21068/AutoCodeAI. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_13120 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | AgentForge: Execution-Grounded Multi-Agent LLM Framework for Autonomous Software Engineering Kumar, Rajesh Ali, Waqar Ahmed, Junaid Ali, Najma Imtiaz Usman, Shaban Software Engineering Artificial Intelligence Large language models generate plausible code but cannot verify correctness. Existing multi-agent systems simulate execution or leave verification optional. We introduce execution-grounded verification as a first-class principle: every code change must survive sandboxed execution before propagation. We instantiate this principle in AGENTFORGE, a multi-agent framework where Planner, Coder, Tester, Debugger, and Critic agents coordinate through shared memory and a mandatory Docker sandbox. We formalize software engineering with LLMs as an iterative decision process over repository states, where execution feedback provides a stronger supervision signal than next-token likelihood. AGENTFORGE achieves 40.0\% resolution on SWE-BENCH Lite, outperforming single-agent baselines by 26--28 points. Ablations confirm that execution feedback and role decomposition each independently drive performance. The framework is open-source at https://github.com/raja21068/AutoCodeAI. |
| title | AgentForge: Execution-Grounded Multi-Agent LLM Framework for Autonomous Software Engineering |
| topic | Software Engineering Artificial Intelligence |
| url | https://arxiv.org/abs/2604.13120 |