Archify Hits GitHub Trending as a Mermaid Alternative That Generates Cleaner Diagrams from Plain English

On June 30, my open-source project Archify appeared on GitHub Trending for the first time, ranking 9th. As of this post, the repository has 4.7k Stars and 414 Forks.

https://github.com/tt-a1i/archify

What is Archify

Archify is an Agent Skill for Claude, Codex, and opencode. You describe a system or process in natural language, and it generates a JSON IR. After schema and layout checks, it outputs a standalone HTML file that supports dark and light themes, copying, and export to PNG, JPEG, WebP, and SVG.

Trending didn't happen overnight

The repository was created on April 15. It started with only architecture diagrams, then gradually added four chart types, dual-theme SVG, 4x resolution export, JSON Schema, layout checks, CLI, CI, and a zero-dependency installation package.

This project gained traction organically and was shared by many bloggers on Twitter. I think the main reasons for its spread are that the generated diagrams genuinely look better than Mermaid, the HTML is easy to share, and the hero image is clear and immediately shows what it does.

Trendshift records show that Archify first entered GitHub Trending on June 30, ranking 9th.

Talking about open source

Let me first talk about my previous experience:

I took the college entrance exam in 2018. I'm not very smart and didn't study well, so I went to a vocational college, majoring in Computer Application Technology. In the first year, I felt very demoralized and couldn't understand the material. After some painful self-reflection, I signed up for the military and served for two years.

When I first joined, my physical fitness was poor. I was thin and lacked strength, weighing only 106 jin, and struggled with many training exercises. Later, I was assigned to a communications unit, which was relatively easier. I trained seriously, and through some opportunities, I represented my unit in a group army communications competition and won first place. I was truly full of pride then, but I won't say more about that.

After leaving the military, I returned to school to continue my studies. Since veterans can enter without an exam, and I didn't know what else to do, I pursued a top-up bachelor's degree. I did fairly well on the exam; out of over seventy people in my cohort, only six were admitted, and I ranked first in the veteran batch. I got into Chongqing University of Posts and Telecommunications, majoring in Software Engineering. I am very grateful to my alma maters, including my vocational college. I feel they took me in, and thanks to them, I was able to see a world beyond that of an average rural kid with mediocre grades.

Because the top-up bachelor's program is only two years, time was tight. At first, I thought about taking the postgraduate entrance exam, but later felt my academic performance wasn't good enough. I was already a vocational student and a veteran, three years older than my peers graduating, so I decided not to.

Then I started learning front-end development, following online courses for over a month, like Shang Silicon Valley. I was quite lucky finding an internship and got one after just over a month of study.

That was in April 2025, at a mid-sized company in Suzhou called PatSnap. The pay was 200 yuan a day, and my rent was 850 yuan. I remember I could still save over two thousand yuan a month. The only hassle was going back to school for exams, a round trip from Suzhou to Chongqing that took over twenty hours and cost nearly two thousand yuan in travel.

But that period was actually quite happy. I was learning things and writing code every day, feeling that as long as I kept working hard, things would definitely get better.

In my fourth month of the internship, I interviewed for ByteDance's Feishu department in Hangzhou. I passed both rounds of interviews, but during the final background check, they found out I was a top-up bachelor's student and blocked me.

At the time, the HR called me, and I kept explaining that I could definitely meet their requirements and prove myself through my work, but in the end, they didn't give me a chance.

That really hurt. But I didn't entirely blame my educational background; I felt more that I just wasn't strong enough. I thought that if my skills were truly exceptional, maybe they wouldn't have blocked me for that reason.

Later, during the autumn recruitment season back at school, I interviewed for Dongchedi in Chongqing. The earlier interviews went fairly well, but when it came to the HR round, I mentioned in advance that I was a top-up bachelor's student, and then there was no follow-up.

The first-round interviewer who contacted me later even proactively sent me a message on Boss saying "sorry." I had a good conversation with him in the first round, so the final rejection probably wasn't his decision.

I replied to him with a long message at the time, but I can't remember the specifics now. I just remember going out for a five-kilometer run that night, feeling a real mix of emotions.

Of course, not all subsequent interview failures were due to my top-up bachelor's background. My lack of skill was definitely part of it.

For example, during the fourth round at Yuanfudao, I couldn't solve the coding problem, so failing was normal. I felt my interviews at Qunhe went fairly well, but I still got rejected in the end, and I don't know the specific reason. Often, I could only guess: maybe it was my education, maybe my ability just wasn't enough.

After failing to find a suitable job during autumn recruitment, I went to Hangzhou and interned at Westlake University for a while.

During that HR interview, they also asked about my top-up bachelor's background. I said, "A humble origin is not a disgrace," and in the end, that interview passed.

By March 2026, I still hadn't found a satisfactory job and was starting to panic.

Later, I went to Shanghai to intern at SenseTime, in the Large Model Research Institute. Our group was almost entirely composed of 985 master's graduates; I was the only one from a vocational college who then did a top-up bachelor's. But SenseTime didn't offer conversion to full-time, so I had to keep looking. After that, I also interviewed at Manus.

I originally applied for a front-end position, but the HR seemed to have messed up the role. In the end, I only had one interview, talking for quite a while with their COO. We discussed a lot: he felt my military service and winning first place in a group army competition were significant plus points and liked me quite a bit. But after hearing I was a top-up bachelor's student, he also directly said it would slightly lower his impression.

We also talked about how someone like me, whose educational background isn't an advantage, could prove their ability through other means, like a portfolio or open-source projects.

Actually, I already had that idea at the time. Because my educational background is something I can't change anymore; no matter how much you explain, what others see is still a top-up bachelor's degree.

So after I joined my first startup, I started writing some blog posts, also posted on Juejin, and later gradually began working on my own projects.

I originally went to that company to do front-end work, but because the company was small, I ended up doing backend, algorithms, client site visits, and client training—all of it.

At the time, it felt like the work was very fragmented, having to do everything. Looking back now, I don't think it was necessarily a bad thing. If I had actually gone to a big company back then, I might have just been doing front-end requirements every day and never been exposed to so many things.

During that period, I also started using AI heavily, made some Claude Code-related tools, and open-sourced some of them.

Honestly, when I first started doing open source, my motives weren't that noble. It was simply because my educational background was poor, and I wanted more things on my resume to prove myself. I still really wanted to get into a big company at that time.

Later, I went to work at MiroFish under Shanda. Outside of work there, I made a lot of my own things, and Archify was started during that period.

After that, I interviewed at a large company in Shanghai—I won't disclose the specifics here. In the first round, there were two coding problems I couldn't solve, and I said I was 100% AI code.

I thought I might fail, but the interviewer then asked me to introduce my projects. I talked about quite a few open-source projects and tools I had built, and in the end, he still let me pass. I haven't been employed there for six months yet, so I won't go into detail about this part.

At the end of June, Archify hit GitHub Trending, peaking at 9th place. Now it has about 4.6k Stars.

Of course, I'm very happy about this data. After all, when I first started, I never expected so many people would pay attention.

But the biggest change after doing open source might be that people have genuinely started using what I've made. Just like in my conversation with the COO of Manus, I now have other things to prove and demonstrate my ability.

Open source hasn't turned me from a top-up bachelor's graduate into a 985 graduate, nor can it guarantee I won't be blocked by my educational background in the future.

I still can't say for sure what these 4.6k Stars will ultimately bring me. Maybe job opportunities, maybe meeting some people, maybe nothing at all.

But at least people are actually using Archify. What I've made can bring value to others and help many people. For me, that is the greatest meaning of doing open source.