diff --git a/src/contract/result.ts b/src/contract/result.ts
index 71f0837..34a81fe 100644
--- a/src/contract/result.ts
+++ b/src/contract/result.ts
@@ -52,6 +52,13 @@ export type ResultNode =
   | { kind: 'gua'; lines: GuaLine[]; title?: string; subtitle?: string }
   /** 文本行（断语、解释） */
   | { kind: 'text'; lines: string[]; lead?: boolean }
+  /**
+   * 可核出处：引文原样列出，并给一个"查原文"的入口。
+   *
+   * 权威性不是靠排版出来的——读者点进去必须在阅览室搜到这句话，
+   * 所以这一节单独成节点，而不是塞进 text 里。
+   */
+  | { kind: 'quote'; book: string; loc?: string; text: string; gloss?: string }
   /** 一句话提示 / 免责 */
   | { kind: 'note'; text: string; tone?: Tone }
 
diff --git a/src/engine/resonance.ts b/src/engine/resonance.ts
index b3b2a0e..cf08297 100644
--- a/src/engine/resonance.ts
+++ b/src/engine/resonance.ts
@@ -105,11 +105,20 @@ function statementNode(m: Matched, index: number): ResultNode {
     }
 
   const s = st.source
-  const lines =
-    s.kind === 'canon'
-      ? [`《${s.book ?? '佚'}》${s.loc ?? ''}`, `「${s.quote ?? ''}」`, s.gloss ?? '']
-      : [s.kind === 'psych' ? '现代心理观察' : '经验观察', s.gloss ?? '']
-  children.push({ kind: 'text', lines: lines.filter(Boolean) })
+  if (s.kind === 'canon' && s.quote) {
+    // 出处必须是可点开的引文节点，不是排好版的字：读者点进去要真能搜到
+    children.push({
+      kind: 'quote',
+      book: s.book ?? '佚',
+      text: s.quote,
+      ...(s.loc ? { loc: s.loc } : {}),
+      ...(s.gloss ? { gloss: s.gloss } : {}),
+    })
+  } else {
+    const lines =
+      s.kind === 'psych' ? ['现代心理观察', s.gloss ?? ''] : ['经验观察', s.gloss ?? '']
+    children.push({ kind: 'text', lines: lines.filter(Boolean) })
+  }
 
   if (st.action) children.push({ kind: 'note', text: '可以做的：' + st.action, tone: 'good' })
 
diff --git a/src/pages/Home.vue b/src/pages/Home.vue
index 8280d1b..2c40312 100644
--- a/src/pages/Home.vue
+++ b/src/pages/Home.vue
@@ -10,6 +10,8 @@ import { CATEGORIES, TOOLS } from '@/tools/registry'
 import { getManifest, type Manifest } from '@/data/corpus'
 import { STATEMENTS } from '@/engine/resonance'
 import { SITUATION_LABEL, SITUATIONS } from '@/engine/features'
+import type { ResultNode } from '@/contract/result'
+import ResultView from '@/ui/ResultView.vue'
 
 const manifest = ref<Manifest | null>(null)
 onMounted(async () => {
@@ -34,6 +36,21 @@ const daily = computed(() => {
 })
 
 const sitChips = SITUATIONS.slice(0, 9).map((s) => SITUATION_LABEL[s])
+
+/** 今日一签的出处也走同一个 quote 节点：能点开核对才算出处，不然只是排版 */
+const dailySource = computed<ResultNode | null>(() => {
+  const s = daily.value.source
+  if (!s) return null
+  if (s.kind === 'canon' && s.quote)
+    return {
+      kind: 'quote',
+      book: s.book ?? '佚',
+      text: s.quote,
+      ...(s.loc ? { loc: s.loc } : {}),
+      ...(s.gloss ? { gloss: s.gloss } : {}),
+    }
+  return { kind: 'text', lines: [s.gloss ?? '经验观察'] }
+})
 </script>
 
 <template>
@@ -46,13 +63,8 @@ const sitChips = SITUATIONS.slice(0, 9).map((s) => SITUATION_LABEL[s])
     <div class="hm-daily">
       <p class="hm-daily-k">今日一签</p>
       <p class="hm-daily-x">{{ daily.text }}</p>
-      <p class="hm-daily-s">
-        <span v-if="daily.source?.kind === 'canon'">
-          《{{ daily.source.book }}》{{ daily.source.loc }} 「{{ daily.source.quote }}」
-        </span>
-        <span v-else>{{ daily.source?.gloss }}</span>
-        · 人群适用约 {{ Math.round(daily.baseRate * 100) }}%
-      </p>
+      <ResultView v-if="dailySource" class="hm-daily-s" :node="dailySource" />
+      <p class="hm-daily-b">人群适用约 {{ Math.round(daily.baseRate * 100) }}%</p>
     </div>
   </section>
 
@@ -111,6 +123,11 @@ const sitChips = SITUATIONS.slice(0, 9).map((s) => SITUATION_LABEL[s])
   border-radius: var(--radius);
   background: var(--bg-1);
 }
+.hm-daily-b {
+  margin: 8px 0 0;
+  font-size: 11px;
+  color: var(--ink-3);
+}
 .hm-daily-k {
   margin: 0 0 6px;
   font-size: 11px;
diff --git a/src/pages/Reading.vue b/src/pages/Reading.vue
index 83c82be..9012015 100644
--- a/src/pages/Reading.vue
+++ b/src/pages/Reading.vue
@@ -5,10 +5,12 @@
  * 参考书要能公开浏览——断语的权威性只有回到原文才成立。
  * 正文不进 bundle：选定书目才 fetch，一部一部懒取。
  */
-import { computed, onMounted, ref, watch } from 'vue'
+import { computed, nextTick, onMounted, ref, watch } from 'vue'
+import { useRoute } from 'vue-router'
 import {
   getChapter,
   getManifest,
+  resolveBooks,
   searchCorpus,
   simplifyAsync,
   type BookMeta,
@@ -19,6 +21,8 @@ import {
 
 const props = defineProps<{ book?: string }>()
 
+const route = useRoute()
+
 const manifest = ref<Manifest | null>(null)
 const error = ref('')
 const query = ref('')
@@ -33,12 +37,73 @@ const toSimplified = ref(false)
 onMounted(async () => {
   try {
     manifest.value = await getManifest()
-    if (props.book) await openBook(props.book)
+    const q = str(route.query.q)
+    const bk = str(route.query.bk)
+    if (q || bk) await openCitation(bk, q)
+    else if (props.book) await openBook(props.book)
   } catch (e) {
     error.value = `语料未就绪：${(e as Error).message}。（部署前需跑 node scripts/sync-corpus.mjs）`
   }
 })
 
+const citation = ref('')
+
+const str = (v: unknown): string => (typeof v === 'string' ? v : '')
+
+/** 由出处跳转写入 query 时，不要惊动输入框那个防抖检索 */
+let suppressQueryWatch = false
+
+/**
+ * 从断语的"查原文"进来：带的是引用书名（可能是《黄帝内经》这种合称）与一段引文。
+ * 先按别名解析站内书目，再在这些书里搜引文，命中就直达那一章。
+ * 搜不到要如实说这本书里没这句话，不能悄悄全域乱搜出一个像样的结果。
+ */
+async function openCitation(citedTitle: string, quote: string) {
+  const m = manifest.value
+  if (!m) return
+  if (!citedTitle) {
+    if (quote) await runSearch(quote)
+    return
+  }
+  const books = resolveBooks(m, citedTitle)
+  if (!books.length) {
+    error.value = `站内没有《${citedTitle}》——这条出处暂时无法核对（需要补书，或在 book-aliases.json 里给别名）。`
+    return
+  }
+  citation.value = `${citedTitle} → ${books.map((b) => `《${b.title}》`).join('、')}`
+  if (!quote) {
+    await openBook(books[0]!.id)
+    return
+  }
+  // 只列命中（含章节与高亮片段），点一下才进正文：
+  // 一进来就跳章会把"这句话到底在不在"的判断权交给页面，读者看不到核的过程。
+  await runSearch(quote, books.map((b) => b.id))
+}
+
+/** 程序化检索：写入输入框、限定书目、拿命中 */
+async function runSearch(q: string, bookIds?: string[]) {
+  clearTimeout(timer)
+  suppressQueryWatch = true
+  query.value = q
+  hits.value = await searchCorpus(q, bookIds, 5)
+  await nextTick()
+  suppressQueryWatch = false
+}
+
+watch(
+  () => [route.query.q, route.query.bk],
+  async ([q, bk]) => {
+    if (!manifest.value) return
+    const quote = str(q)
+    const title = str(bk)
+    if (quote || title) {
+      error.value = ''
+      citation.value = ''
+      await openCitation(title, quote)
+    }
+  },
+)
+
 watch(
   () => props.book,
   (id) => {
@@ -71,6 +136,8 @@ async function openChapter(chapterId: string) {
 let timer: ReturnType<typeof setTimeout> | undefined
 watch(query, (q) => {
   clearTimeout(timer)
+  // 出处跳转自己会把限定书目的命中算好，这里再跑一遍会退化成全域搜索
+  if (suppressQueryWatch) return
   if (q.trim().length < 2) {
     hits.value = null
     return
@@ -158,6 +225,11 @@ async function jumpTo(hit: Hit) {
 
     <p v-if="error" class="rd-err">{{ error }}</p>
 
+    <p v-if="citation" class="rd-cite">
+      正在核对出处：{{ citation }}
+      <span v-if="hits && !hits.length">—— 这句话在本段引文里没搜到，可手动改上面的检索词</span>
+    </p>
+
     <input v-model="query" class="du-input rd-q" placeholder="跨全书检索，例如 恬淡虚无 / 体用 / 顺逆之机" />
 
     <nav v-if="manifest" class="rd-books">
@@ -246,6 +318,12 @@ async function jumpTo(hit: Hit) {
   border: 1px dashed var(--tone-bad);
   border-radius: 6px;
 }
+.rd-cite {
+  margin: 0;
+  font-size: 11px;
+  letter-spacing: 0.04em;
+  color: var(--accent-2);
+}
 .rd-q {
   margin-top: 4px;
 }
diff --git a/src/ui/ResultView.vue b/src/ui/ResultView.vue
index d1ff01c..c99bfb1 100644
--- a/src/ui/ResultView.vue
+++ b/src/ui/ResultView.vue
@@ -10,6 +10,21 @@ import type { MatrixCell, ResultNode } from '@/contract/result'
 
 const props = defineProps<{ node: ResultNode }>()
 
+/**
+ * 出处跳转：带引用书名 + 首段引文进阅览室，由阅览室做书名别名解析与检索。
+ *
+ * 这里不猜站内书目 id（那要 fetch manifest，渲染器不该有 IO），也不装 RouterLink
+ * （ResultView 是递归组件，被单测和首页独立挂载时没有 router 实例）：
+ * hash 路由下 `#/reading?...` 就是一个合法的相对链接，两个平台同一份产物。
+ */
+const quoteHref = computed(() => {
+  if (props.node.kind !== 'quote') return null
+  const seg = props.node.text.split(/…+/)[0]!.trim()
+  const q = new URLSearchParams({ bk: props.node.book, q: seg })
+  if (props.node.loc) q.set('loc', props.node.loc)
+  return `#/reading?${q.toString()}`
+})
+
 /** 卦画：上爻显示在最上 */
 const guaRows = computed(() =>
   props.node.kind === 'gua' ? [...props.node.lines].reverse() : [],
@@ -143,6 +158,15 @@ function widthOf(value: number, max?: number): string {
     <p v-for="(l, i) in node.lines" :key="i">{{ l }}</p>
   </div>
 
+  <div v-else-if="node.kind === 'quote'" class="qt">
+    <p class="qt-x">「{{ node.text }}」</p>
+    <p class="qt-s">
+      <span class="qt-b">《{{ node.book }}》{{ node.loc ?? '' }}</span>
+      <span v-if="node.gloss" class="qt-g">{{ node.gloss }}</span>
+      <a v-if="quoteHref" class="qt-a" :href="quoteHref">查原文</a>
+    </p>
+  </div>
+
   <p v-else-if="node.kind === 'note'" class="nt" :class="toneClass(node.tone)">{{ node.text }}</p>
 </template>
 
@@ -452,4 +476,35 @@ function widthOf(value: number, max?: number): string {
   border: 1px dashed var(--line);
   color: var(--ink-3);
 }
+.qt {
+  display: grid;
+  gap: 4px;
+  padding: 10px 12px;
+  border-left: 2px solid var(--accent-2);
+  background: var(--surface-2);
+  border-radius: 0 6px 6px 0;
+}
+.qt-x {
+  margin: 0;
+  font-size: 13px;
+  line-height: 1.9;
+  color: var(--ink-1);
+}
+.qt-s {
+  display: flex;
+  flex-wrap: wrap;
+  align-items: baseline;
+  gap: 8px;
+  margin: 0;
+  font-size: 11px;
+  color: var(--ink-3);
+}
+.qt-a {
+  margin-left: auto;
+  padding: 2px 9px;
+  border: 1px solid var(--accent-2);
+  border-radius: 999px;
+  color: var(--accent-2);
+  text-decoration: none;
+}
 </style>
diff --git a/tests/app.spec.ts b/tests/app.spec.ts
index 6e282dc..8a390ba 100644
--- a/tests/app.spec.ts
+++ b/tests/app.spec.ts
@@ -61,8 +61,10 @@ describe('界面冒烟', () => {
     await flushPromises()
     expect(w.text()).toContain('梅花易数')
     expect(w.findAll('.hm-card')).toHaveLength(1)
-    // 每日一签必须带出处或释义，否则等于匿名断语
+    // 每日一签必须带出处或释义，否则等于匿名断语；canon 的话还得能点开核对
     expect(w.find('.hm-daily-s').text().length).toBeGreaterThan(4)
+    const dailyLink = w.find('.hm-daily-s .qt-a')
+    if (dailyLink.exists()) expect(dailyLink.attributes('href')).toContain('#/reading')
   })
 
   it('工具页：填领域 → 点起一卦 → 出卦画、断语与反馈区', async () => {
@@ -128,6 +130,7 @@ describe('界面冒烟', () => {
   it('ResultView 能解释全部节点类型（新工具不需要新组件）', () => {
     const nodes: ResultNode[] = [
       { kind: 'text', lines: ['x'], lead: true },
+      { kind: 'quote', book: '黄帝内经', loc: '素问 · 四气调神大论', text: '不治已病治未病', gloss: '先防' },
       { kind: 'note', text: 'n', tone: 'warn' },
       { kind: 'chips', items: [{ text: 'c', tone: 'good' }] },
       { kind: 'kv', items: [{ label: 'a', value: 'b', sub: 'c' }] },
@@ -145,10 +148,31 @@ describe('界面冒烟', () => {
       { kind: 'section', title: 'S', children: [{ kind: 'note', text: 'in' }] },
     ]
     const w = mount(ResultView, { props: { node: { kind: 'stack', children: nodes } } })
-    for (const cls of ['.tx', '.nt', '.ch', '.kv', '.gr', '.br', '.sco', '.tb', '.mx', '.ga', '.rw', '.sc'])
+    for (const cls of [
+      '.tx',
+      '.nt',
+      '.ch',
+      '.kv',
+      '.gr',
+      '.br',
+      '.sco',
+      '.tb',
+      '.mx',
+      '.ga',
+      '.rw',
+      '.sc',
+      '.qt',
+    ])
       expect(w.find(cls).exists(), `节点未渲染：${cls}`).toBe(true)
     // 环形评分的描边长度必须随分值变化，否则是摆设
     expect(w.find('.sco-fg').attributes('stroke-dasharray')).toBe('215.6 326.7')
+    // 出处的"查原文"必须把书名与引文带进阅览室查询串，别名交给阅览室解析
+    const href = w.find('.qt-a').attributes('href')!
+    expect(href).toContain('#/reading')
+    const q = new URLSearchParams(href.split('?')[1]!)
+    expect(q.get('bk')).toBe('黄帝内经')
+    expect(q.get('q')).toBe('不治已病治未病')
+    expect(q.get('loc')).toBe('素问 · 四气调神大论')
   })
 
   it('FormRenderer 按 schema 生成控件，必填未填时提示', async () => {
diff --git a/tests/citation.spec.ts b/tests/citation.spec.ts
new file mode 100644
index 0000000..b5b154d
--- /dev/null
+++ b/tests/citation.spec.ts
@@ -0,0 +1,87 @@
+/**
+ * 出处闭环
+ *
+ * 断语上那句"查原文"是唯一能让权威性当场兑现的入口：
+ * 读者点进去必须在阅览室搜到这句话。这条链路跨了三个模块
+ * （引擎出 quote 节点 → ResultView 拼查询串 → 阅览室解析书名别名并检索），
+ * 任一处写歪，用户看到的就是一句无处可核的引文。
+ */
+import { readFileSync } from 'node:fs'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { createMemoryHistory } from 'vue-router'
+import { flushPromises, mount } from '@vue/test-utils'
+import { makeRouter } from '@/router'
+import Reading from '@/pages/Reading.vue'
+import ResultView from '@/ui/ResultView.vue'
+import { STATEMENTS } from '@/engine/resonance'
+import type { ResultNode } from '@/contract/result'
+
+/** 与 corpus.spec 同一口径：读 public/corpus，即部署产物本身 */
+async function fakeFetch(input: string) {
+  const path = input.replace(/^(\.\/|\/)/, '')
+  try {
+    const body = JSON.parse(readFileSync(`public/${path}`, 'utf8'))
+    return { ok: true, json: async () => body } as unknown as Response
+  } catch {
+    return { ok: false, status: 404, json: async () => ({}) } as unknown as Response
+  }
+}
+
+async function settle(rounds = 8): Promise<void> {
+  for (let i = 0; i < rounds; i++) {
+    await flushPromises()
+    await new Promise((r) => setTimeout(r, 0))
+  }
+}
+
+beforeEach(() => {
+  vi.stubGlobal('fetch', vi.fn(fakeFetch))
+})
+
+describe('断语 → 原文 的闭环', () => {
+  it('每条 canon 断语都渲染成带查询串的出处链接', () => {
+    const canon = STATEMENTS.filter((s) => s.source?.kind === 'canon' && s.source.quote)
+    expect(canon.length).toBeGreaterThanOrEqual(20)
+    for (const st of canon) {
+      const node: ResultNode = {
+        kind: 'quote',
+        book: st.source!.book!,
+        text: st.source!.quote!,
+        ...(st.source!.loc ? { loc: st.source!.loc } : {}),
+      }
+      const w = mount(ResultView, { props: { node } })
+      const href = w.find('.qt-a').attributes('href')
+      expect(href, `${st.id} 的出处不可点`).toContain('#/reading')
+      const q = new URLSearchParams(href!.split('?')[1]!)
+      // 引文进查询串时必须是去省略号后的首段，否则阅览室永远搜不到
+      expect(q.get('q')!.length).toBeGreaterThan(3)
+      expect(q.get('q')).not.toContain('…')
+    }
+  })
+
+  it('点进阅览室：《黄帝内经》按别名解析到素问/灵枢，并搜出那句话', async () => {
+    const router = makeRouter(createMemoryHistory())
+    await router.push({
+      name: 'reading',
+      query: { bk: '黄帝内经', q: '恬淡虛無，真氣從之' },
+    })
+    await router.isReady()
+    const w = mount(Reading, { global: { plugins: [router] } })
+    await settle(30) // 首次要载 opencc + 拉两部书的 JSON
+
+    expect(w.find('.rd-cite').text()).toContain('素问')
+    expect(w.findAll('.rd-hit').length, '引文在站内搜不到').toBeGreaterThan(0)
+    expect(w.find('.rd-hit').text()).toContain('恬淡虚无') // snippet 已转简体
+  })
+
+  it('书名对不上时如实说无法核对，不假装命中', async () => {
+    const router = makeRouter(createMemoryHistory())
+    await router.push({ name: 'reading', query: { bk: '不存在的书', q: '随便一句' } })
+    await router.isReady()
+    const w = mount(Reading, { global: { plugins: [router] } })
+    await settle()
+
+    expect(w.find('.rd-err').text()).toContain('站内没有《不存在的书》')
+    expect(w.findAll('.rd-hit').length).toBe(0)
+  })
+})
diff --git a/tests/tool.spec.ts b/tests/tool.spec.ts
index 4c3b428..830ec04 100644
--- a/tests/tool.spec.ts
+++ b/tests/tool.spec.ts
@@ -23,6 +23,7 @@ const NODE_KINDS = [
   'matrix',
   'gua',
   'text',
+  'quote',
   'note',
 ] as const
 const TONES: Tone[] = [
@@ -150,13 +151,30 @@ describe('工具输出结构', () => {
     const out = await cases[0]!.tool.compute(cases[0]!.input)
     if (!('body' in out)) throw new Error('no body')
     const texts = collectNodes(out.body)
-      .filter((n) => n.kind === 'note' || n.kind === 'text')
-      .map((n) => (n.kind === 'note' ? n.text : n.lines.join('')))
+      .filter((n) => n.kind === 'note' || n.kind === 'text' || n.kind === 'quote')
+      .map((n) =>
+        n.kind === 'note' ? n.text : n.kind === 'text' ? n.lines.join('') : `《${n.book}》${n.text}`,
+      )
       .join('\n')
     expect(texts).toContain('梅花易数')
     expect(texts).toMatch(/不作医疗|不作为医疗/)
   })
 
+  it('canon 出必须是 quote 节点（可点开核对），不许退化成排好版的纯文本', async () => {
+    const out = await cases[0]!.tool.compute(cases[0]!.input)
+    if (!('body' in out)) throw new Error('no body')
+    const quotes = collectNodes(out.body).filter((n) => n.kind === 'quote')
+    // 至少要有 canon 出现，否则这条断言是空过的
+    expect(quotes.length, '结果里一条 canon 出处都没有').toBeGreaterThan(0)
+    for (const q of quotes) {
+      if (q.kind !== 'quote') continue
+      expect(q.book.length, '出处没书名').toBeGreaterThan(0)
+      expect(q.text.length, '出处没引文').toBeGreaterThan(3)
+      // 引文不该被截断成省略号尾巴——截断就是凭印象引书的痕迹
+      expect(q.text).not.toMatch(/…$/)
+    }
+  })
+
   it('非法输入返回 error 而不是抛异常或空结果', async () => {
     const t = toolById('meihua')!
     const bad = await t.compute({ mode: 'number', numA: 0, numB: 1, domain: 'career' })