AgileCoder: Dynamic Collaborative Agents for Software Development based on Agile Methodology
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Software agents are emerging as powerful tools for tackling complex software engineering tasks. However, existing approaches often oversimplify development workflows, assuming basic models that lack the nuances of real-world software processes. Moreover, they frequently include entire codebases in their instructions, leading to inefficiencies when working with large-scale projects. To overcome these challenges, we introduce AgileCoder, a multi-agent system that incorporates Agile Methodology (AM) principles, assigning specialized agents to roles such as Product Manager, Developer, and Tester for collaborative, iterative development. AgileCoder structures work into sprints, enabling incremental progress based on user input. A standout feature, the Dynamic Code Graph Generator, continuously builds a Code Dependency Graph as the codebase evolves, allowing agents to gain a deeper understanding of the structure for more precise code generation and efficient modifications. We evaluate AgileCoder on two fronts: (1) code generation benchmarks, including HumanEval and MBPP, and (2) real-world software development scenarios. The results show that AgileCoder outperforms existing systems like ChatDev and MetaGPT, setting a new standard for multi-agent systems in software engineering. The source code is available at https://github.com/FSoft-AI4Code/AgileCoder.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it