f2d7087c21
同事提供了正确的速卖通链接，替换掉 babel fuzzy 自动乱配的那批 （原先多个不同淘宝 ID 被配到同一个速卖通地址，明显是错配， 当时保留 fuzzy 让它回退到淘宝链接，没有发布出去）。 SLogic16U3 / MaixCAM2 / MaixCAM / Maix-IV NanoKVM PCIe / Pro / USB Tang Mega Console / Tang Nano 4K 9 条全部填入并清除 fuzzy 标记，构建后英文首页确认已跳速卖通。 仍跳淘宝的 3 个产品（T256s、Maix-I K210、Maix-III AXeraPi）本就 没有速卖通译文，不在这次的错配范围内，待有链接再补。
Sipeed wiki source docs
Sipeed official Wiki: wiki.sipeed.com
Contributing and article sharing
You are welcome to contribute to the wiki, fixing error, adding content, or sharing your article are all welcome.
Fork this repo and change docs, then create a Pull Request.
More contribute doc :
Preview locally
git clone --depth=1 https://github.com/sipeed/sipeed_wiki.git
cd sipeed_wiki
uv sync
uv run teedoc serve
uv sync creates/updates the .venv from pyproject.toml and uv.lock, which already includes teedoc and all plugins used by this site (so teedoc install is not needed).
Install uv first if you don't have it.
Why --depth=1
This is a documentation site, so most of the repo is images. Git keeps every past revision of every one of them forever, including large originals that have since been replaced by compressed versions. A full clone therefore downloads roughly 1 GB of history even though you only ever need the current version of each file.
--depth=1 fetches just the latest commit and skips all of that. Editing docs and opening
a pull request works exactly the same from a shallow clone. Use a full clone only when you
actually need the history (git log, git blame, git bisect) — you can always deepen a
shallow clone later with git fetch --unshallow.
When adding images, please keep them small (ideally under 300 KB, 500 KB max): anything committed here stays in the history for good, so an oversized image costs every future contributor bandwidth even after it is deleted. See AGENTS.md for the full image and authoring guidelines.
More build tool usage see teedoc
For administrator
GitHub action will automatically push changed doc to cloud server. But if you did some special things like force commit, some files maybe missing on server, manually trigger the publish_upload_all action to build doc and upload all files to server.