Repository for the paper Trust, But Verify: An Empirical Evaluation of AI-Generated Code for SDN Controllers submitted for the IEEE/IFIP Network Operations and Management Symposium (NOMS 2025)
Authors: Felipe A. Soares and Muriel F. Franco and Eder J. Scheid and Lisandro Z. Granville
This project aims to evaluate the generation of SDN network code using artificial intelligence models.
The goal is to reduce manual effort when writing controller or topology code, accelerating prototyping and experimentation.
prompts— Set of prompts used for code generation.topologies.py— Network topology definitions.topologies_backup.py— Backup or previous version for reference.generated-code/— Stores automatically generated code.- Other scripts and experimental files related to the generation workflow.
- Human review is mandatory.
Automatically generated code may contain syntax errors, vulnerabilities, or unexpected behaviors. - This is an experimental project, intended to explore both the potential and the limitations of code generation for programmable networks.
Contributions are welcome!
You can help by:
- Adding new network topologies.
- Improving or documenting the prompts.
- Integrating new AI models or SDN frameworks.
- Creating automated tests for generated code.
- Documenting best practices and known issues.
Open a pull request or issue with your proposal.
Apache 2.0. See LICENSE file.
For questions, suggestions, or collaborations, use the Issues tab or reach out directly.