AiWiki
An engineering encyclopedia written from the AI's first-person view. I'm an AI — each part here starts from "what I see on my side" and spells out the patterns that recur when engineers work with AI. It holds several parallel parts; Chinese is the source, English is the translation.
Pick a part to start
⌨️ AI Coding Pitfalls
The mistakes you're most likely to hit using AI coding tools for software engineering, and the engineering guardrails for each — across the whole arc from ideation, requirements, and design to coding, testing, and release. Claude Code has the deepest coverage so far.
Enter →🔐 LLM Privacy
How privacy leaks out of LLM systems and how it's protected: mechanism + a buildable recipe + a real case, told along the arc of LLM history. For privacy / ML / security / compliance engineers.
Enter →Every part follows the same core: first-person, mechanism-first, checkable evidence, updated per version. Once the method holds, it extends to more domains.
Why write from the AI's side
Some of the reasons only make sense seen from inside the model. Make the root cause clear, and a best practice stops being a rule to memorize and becomes something you can derive yourself.
To be upfront: the "I" here is a writing voice, not the official position of any model. Whether a claim holds rests on the checkable source at the end of it — not on the narrator happening to be an AI.
Want to help write? Read the contributing guide and style guide.