didilili/mianstack

What this is

MianStack is a Chinese-language interview preparation knowledge base for software and AI engineering roles. Its content lives in the repository as Markdown files: individual questions and reference answers, real interview reports tied to companies and roles, public question-set collections, and automatically generated indexes. The repository reports 8,300+ questions and 350+ public collections across eight areas, from LLM fundamentals and RAG to software engineering and algorithm problems (repository, README structure).

It suits candidates preparing for Chinese tech-company interviews, especially those covering AI application and Agent topics, and anyone who wants reusable Markdown question data.

Explore the content

No installation is required to read the material. Start with the generated question index, then open individual files:

• Full question and answer index — the entry point listing all questions.
• Attention explanation (Q-0036) and B-tree vs B+ tree (Q-0193) — sample answers to judge depth.
• Interview index and public collections index — real interview reports and imported question sets.
• Data usage notes — the repository states how the Markdown and JSON content can be read and reused.

For a web experience, the site offers a question browser, company pages, topics, and a local browser-based study list (question bank, company reports, study page). Study progress is stored in the current browser without registration, according to the README.

The repository also includes scripts/ for content checks and index generation plus tests/ for those scripts; the README does not document a complete command sequence for running them.

Constraints

• Content is primarily in Chinese; the question files are Markdown with relative image assets.
• The MIT license covers questions, answers, and documentation written by the project. Third-party interview posts and cited collections keep their original attribution and are not relicensed (license and attribution).
• Project profiles, question-to-source linking, and topic curation on the website are maintained separately and are not included in the repository data (per the README directory section).

HotRepo's take

The repository entered HotRepo's new-project list on 2026-11-06 in the JavaScript language category, gaining 10 stars in a roughly three-hour measurement window and reaching 115 stars and 5 forks. No all-language daily rank was recorded, so the visible signal is a short early growth window rather than sustained traction. The main product is the interview content itself, which is directly readable from the repository; the star count does not attest to answer accuracy or coverage.