Japan’s Digital Agency Releases Harunobu

Japan’s Digital Agency released Harunobu on October 21, 2026, to assess whether files such as spreadsheets are easy for computers to read. The agency says the tool can help improve files known as “God Excel,” which often prioritize appearance on printed pages and use features such as merged cells. Files used by government and businesses are not always easy for data systems or AI to handle, so the tool is intended to support data preparation.

The agency developed Harunobu and published it on GitHub. It scores Excel, CSV and TSV files out of 100 according to rules for making government data easier for systems and AI to use. Its rule-based checks avoid the probabilistic errors that can occur with generative AI and run relatively lightly. The agency says it can process files faster than people.

Harunobu outputs scores in JSON, allowing them to be used with external systems, including generative AI services. A Digital Agency blog showed an example in which ChatGPT uses the scores to create a program that modifies Excel files and prepares the data. The agency expects the tool to help reduce work time and support data preparation.

The name Harunobu comes from Suzuki Harunobu, an ukiyo-e artist of the Edo period. Nishiki-e, whose development Harunobu is said to have played a central role in, are made by layering colors from multiple printing blocks on one sheet of paper. The agency said it chose the name to connect that layering process with applying multiple machine-readability rules to “God Excel” files to create data that computers can read. Machine readability is the quality that lets computers read and process data. JSON is a format for transferring data between systems; CSV separates fields with delimiters, while TSV separates them with tabs. The article gives no details of the scoring rules or the actual time saved. In the example, ChatGPT creates an Excel-processing program based on the scores, and the agency says Harunobu is intended to reduce work time and support data preparation across government and businesses.

Japan’s Digital Agency released Harunobu on October 21, 2026, to assess whether files such as spreadsheets are easy for computers to read. The agency says the tool can help improve files known as “God Excel,” which often prioritize appearance on printed pages and use features such as merged cells. Files used by government and businesses are not always easy for data systems or AI to handle, so the tool is intended to support data preparation.

The agency developed Harunobu and published it on GitHub. It scores Excel, CSV and TSV files out of 100 according to rules for making government data easier for systems and AI to use. Its rule-based checks avoid the probabilistic errors that can occur with generative AI and run relatively lightly. The agency says it can process files faster than people.

Harunobu outputs scores in JSON, allowing them to be used with external systems, including generative AI services. A Digital Agency blog showed an example in which ChatGPT uses the scores to create a program that modifies Excel files and prepares the data. The agency expects the tool to help reduce work time and support data preparation.

The name Harunobu comes from Suzuki Harunobu, an ukiyo-e artist of the Edo period. Nishiki-e, whose development Harunobu is said to have played a central role in, are made by layering colors from multiple printing blocks on one sheet of paper. The agency said it chose the name to connect that layering process with applying multiple machine-readability rules to “God Excel” files to create data that computers can read. Machine readability is the quality that lets computers read and process data. JSON is a format for transferring data between systems; CSV separates fields with delimiters, while TSV separates them with tabs. The article gives no details of the scoring rules or the actual time saved. In the example, ChatGPT creates an Excel-processing program based on the scores, and the agency says Harunobu is intended to reduce work time and support data preparation across government and businesses.