Worked on the apache/incubator-hugegraph-ai repository to address a critical issue in LLM-driven Gremlin prompt formatting. Focused on improving the reliability of graph analytics by implementing a _format_properties helper in Python, ensuring the 'properties' field is always included in generated prompts. This fix resolved inconsistencies in Gremlin query generation, reducing manual intervention and enhancing maintainability of the prompt engineering codebase. Leveraged skills in LLM integration and prompt engineering to deliver a targeted bug fix that improved the accuracy and completeness of AI-generated queries, directly supporting HugeGraph AI’s business needs for dependable and automated graph data analysis workflows.