This study introduces a novel Large Language Model (LLM)-driven approach for comprehensive diagnosis and prognostics of vehicle faults, leveraging Diagnostic Trouble Codes (DTCs) in line with industry-standard automation protocols. The proposed model asks for significant advancement in automotive diagnostics by reasoning through the root causes behind the fault codes given by DTC document to enhance fault interpretability and maintenance efficiency, primarily for the technician and in few cases, the vehicle owner. Here LLM is trained on vehicle specific service manuals, sensor datasets, historical fault logs, and Original Equipment Manufacturer (OEM)-specific DTC definitions, which leads to context-aware understanding of the vehicle situation and correlation of incoming faults. Approach validation has been done using field level real-world vehicle dataset for different running scenarios, demonstrating model’s ability to detect complex fault chains and successfully predicting the associated root cause. By utilizing time series based future projection of the vehicle pattern, this approach could also predict the probable future faults as well as the requisite steps needed to prevent them. Overall, key contributions of this work include: (1) a modular diagnostic framework that seamlessly integrates different electronic control unit (ECU) architectures for sequential root cause analysis of vehicle faults, (2) cross-platform compatibility allowing utility across varied vehicle models and platforms, and (3) a user-friendly interface that eliminates the need for technical expertise by generating output data into simple, actionable insights. This work was benchmarked against traditional rule-based diagnostic tools and showed 50-70% reduction in the troubleshooting time for Root cause analysis (RCA). In the prognosis front, model could predict upcoming possible faults in the Battery behavior with significant accuracy. The framework also supports continuous learning by integrating new fault patterns, ensuring adaptability over time. This paper establishes the potential of integrating advanced language models into the automotive diagnostics pipeline and provides a scalable, intelligent, and intuitive solution for next-generation vehicle fault management.