Abstract
This document describes a self-improving code generation system, also known as a programming assistant, software synthesis tool, or automated coding assistant, that dynamically enhances automated code generation accuracy by using an artificial intelligence (AI) or machine learning system, such as a large language model (LLM), by continuously learning from user interactions. During application development sessions, the system evaluates user feedback in parallel with code generation tasks to separate subjective personal tastes from objective functional corrections. The system distills and merges functional corrections into an aggregated guidelines document that acts as an evolving set of engineering rules so as to enable non-parametric learning and parallel feedback distillation. Subsequent code generation sessions automatically include the aggregated guidelines document as a supplemental prompt, shadow prompt, or dynamic context injection so as to guide output and perform automated prompt maintenance, allowing the generative AI tool to produce more accurate code without incurring model retraining costs.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Bertran, Ishac, "CODE GENERATION SYSTEM WITH INTEGRATION OF OBJECTIVE USER FEEDBACK", Technical Disclosure Commons, (June 25, 2026)
https://www.tdcommons.org/dpubs_series/10569