MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641126575 A) filed by Dr. Sarojini Y; Raghunatha Reddy T G; Mr. Ayush Kumar Choudhary; Ms. Jyotsna G; and Ms. Aptha K. S. on October 23, 2026, for A Mutation-Informed Trust Evaluation Framework For Autonomous Software Repair And Controlled Patch Acceptance.
Inventors include Dr. Sarojini Y; Raghunatha Reddy T G; Mr. Ayush Kumar Choudhary; Ms. Jyotsna G; and Ms. Aptha K. S..
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: A Mutation-Informed Trust Evaluation Framework for Autonomous Software Repair and Controlled Patch Acceptance ABSTRACT OF THE INVENTION The present invention relates to a computer-implemented autonomous software recovery system and method using artificial-intelligence-based patch synthesis and mutation- based trust evaluation. The system captures runtime failures, performs anomaly triage, localizes faults using traceback and Abstract Syntax Tree information, generates candidate patches using a Large Language Model, and validates the candidates in isolated execution environments. After a candidate passes ordinary tests, the system identifies the changed region and performs controlled AST-level mutation testing. Killed and surviving valid mutants provide evidence of local test sensitivity. The mutation evidence is combined with patch-risk and governance constraints to generate a trust assessment used by a policy engine to select automatic merge, human review, or rejection. The system further supports repository orchestration, auditability, and recovery-memory retrieval. The trust assessment is an acceptance signal and does not constitute proof of semantic correctness.
Disclaimer: Curated by HT Syndication.
