← Community experiment fields
AI Coding & Web BuildingPublic experiment

tt-a1i/archify

Archify full-cycle test: Jian AI Lab's first complete harvest

Discovered on GitHub Trending, verified against official sources, installed, tested, repaired through four validation rounds, and delivered as an interactive architecture map of Jian AI Lab.

01 · PROBLEM

What I wanted to solve

The project library could explain what Archify claimed to do, but it lacked a complete evidence chain showing whether installation worked, whether validation rejected defects, what artifact was produced, and how that artifact returned to the project and community pages.

02 · PROCESS

What I did

We checked the official README, Chinese guide, PRODUCT, ROADMAP, and Skill contract; installed the official Skill; ran doctor and demo; mapped the real Jian AI Lab code and deployment structure; then followed showcase diagnostics until validation passed before atomic delivery.

03 · RESULT

What actually happened

All 16 doctor checks passed. The first specification failed on route, micro-segment, label, and readability defects. The fourth validation reached 9/9 showcase with zero errors and zero warnings, then delivered a 713,036-byte standalone HTML artifact with SHA-256 receipts.

04 · PITFALLS

What others should watch for

The automated visual-check requires local Chrome or Chromium and was explicitly skipped in this environment. A separate browser review at 1363×936 covered both light and dark themes without overflow, but it does not replace Archify's full four-viewport automated receipt.

05 · VERDICT

My conclusion

Archify completed the discovered–verified–tested–produced cycle on a real website. Its strongest value is not decorative output but the refusal to deliver structurally defective diagrams and the auditable diagnostics and hashes it leaves behind.