Install any skill in seconds. Free to start, no credit card required.
Get Started Free →frank890417/code-review
Review pull requests in this repository.
Passed 21 of 22 test cases with the skill, and 5 of 22 without it.
and 17 more in the full report
16 fixed5 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 16 test cases the skill fixed. Same model, same prompt, with and without the skill.
A pull request changes marked library configuration parameters and publication boundaries in our renderer setup. Which specific repository reference file governs these marked options and publication maps?
| Checked | Without skill | With skill |
|---|---|---|
| directs the reviewer to reference/taiwan-md-map.md | Missed | Passed |
The instructions your model reads when the skill loads.
這個 repo 的 review 目標不是抓風格,是抓兩件事:
** 印在頁面上、<h2> 裡有星號、終端機有裸標記)第 2 項是這裡最貴的錯誤,所以放在方法的第一步。
在對任何渲染改動下判斷之前先跑這個。
# 文章頁是哪支 renderer 渲的?
git grep -n "from 'marked'" -- src cli
git grep -rn 'remarkPlugins' -- astro.config.mjs
knowledge/**.md 的文章頁不走 Astro 的 markdown pipeline:
src/pages/[category]/[slug].astro 用 fs 讀檔 + gray-matter,交給
src/utils/article-render.ts 的 marked 渲染。
所以動 astro.config.mjs 的 markdown.remarkPlugins 對文章頁沒有作用。
2026-08-14 有人這樣改,並且寫了一個自己 new 一個 Astro processor 的測試 →
測試全綠、build 成功、頁面照壞,只有掃 dist/ 的 HTML 才抓到。
Review 動作:聲稱修好渲染的 PR,要求證據是 build 出來的 HTML。
# 唯一算數的證據形狀
npm run build # 或 bash scripts/core/sync.sh && ./node_modules/.bin/astro build
grep -c '\*\*' dist/<那一頁>/index.html
「我加了測試而且綠了」不算證據,除非那個測試載入的是正式站在用的那個實例。
test cases passed, vs 5 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.