Having focused on backend development for many years, I recently started exploring AI programming and dabbling in frontend-related content. However, during this experimentation, I’ve gradually realized I’m falling back into an old habit – being distracted by shiny things. I constantly want to use AI to build a frontend interface, but these attempts offer little practical benefit for my current work and instead disperse my focus.

This site is built with Hugo, but I’ve been using Chinese titles, which results in unfriendly URLs. To put it simply, when shared, they don’t look good because Chinese characters are encoded as things like %E4%BD%A0%E5%A5%BD. While slugs can fix this, manually setting them each time is too much trouble.

So, today I tried using Claude4 to develop a translation assistant that automatically converts Chinese titles into English slugs and adds hyperlinks within articles. This would eliminate the need for manual setup.

Claude 4 is excellent; its ability to understand context and handle complex tasks has been significantly improved

China’s family planning policy, while limiting population growth, curbed the development of family-based enterprises, undermined traditional social structures, and suppressed the expansion of family businesses and political influence. Compared to South Korea’s chaebols and India’s family monopolies, its uniqueness is evident. Now that birth restrictions have been lifted, although challenges such as low birth rates are faced, new monopoly risks must also be guarded against, and a balance needs to be sought through multiple avenues.

Since the new AI Inspiration Hub was established, a lot of trivial things have been tried to be recorded and published using AI. As a result, there’s less time for quiet reflection and contemplation. It would be good to slightly control the output of this column in the future and integrate it into a monthly publication format, releasing one article per month.

It’s like a kind of aftereffect, or perhaps a side effect – efficiency is improved, but the depth and breadth of thinking have declined

The evolution of tariffs in the 2025 trade war presented a cycle of “escalation — retaliation — consultation,” as the US-China competition shifted from tariff confrontation to rules-based competition. While short-term easing alleviated market pressure, long-term uncertainty remains, and it is necessary to monitor the continued impact on the global economy from WTO rulings, supply chain adjustments, and geopolitical changes.

People can’t earn money beyond their own understanding. The stock market crash triggered by the trade war this year, even now that losses have been largely recovered, has buried countless small investors along the way.

The 2020-2025 teacher recruitment in Jiangxi Province shows a significant reduction trend: the number of special post teachers recruited decreased sharply from 6,617 to 32 (a decrease of 99.5%), and national cadre teachers decreased from 11,324 to 2,146 (a decrease of 81.1%). The proportion of core subjects (language, mathematics, English) remains stable but the total number is shrinking; although the proportion of subjects such as music, arts, and physical education has increased, the absolute numbers are limited (e.g., only 2 people were recruited in 2025 for each). Policy changes have implemented “one position reduction, one position replacement” staffing restrictions, with teacher resources shifting towards vocational education and remote areas, leading to a significant reduction in traditional primary and secondary school positions. In 2025, the planned number for some subjects was zero.

github-readme-stats is a statistics generator for GitHub profiles, allowing users to display various statistics and charts on their GitHub profile. It offers multiple customization options that can be adjusted according to user needs.

I manage my warehouse by grouping them into projects. GitHub doesn’t support warehouse grouping, so I have to achieve this by splitting them into different organizations. The latest branch of github-readme-stats cannot support statistics across different organizations’ warehouses, so I forked a branch and merged the corresponding code.

The script is used to batch modify the author information of Git history commits by rewriting commit records using git filter-branch