diff --git a/server/routes.ts b/server/routes.ts
index ca74bdd..c626298 100644
--- a/server/routes.ts
+++ b/server/routes.ts
@@ -47,6 +47,8 @@ const WRITABLE: Array<{ method: 'PUT' | 'DELETE'; pattern: RegExp }> = [
   { method: 'DELETE', pattern: /^\/connections(\/[^/]+)?$/ },
   // 让内核重新拉取某个订阅
   { method: 'PUT', pattern: /^\/providers\/proxies\/[^/]+$/ },
+  // 让内核重新拉取某个规则组
+  { method: 'PUT', pattern: /^\/providers\/rules\/[^/]+$/ },
 ]
 
 /** 允许透传的管理接口路径。白名单而不是黑名单——多开一个口子要有意为之。 */
@@ -59,6 +61,8 @@ const READABLE_KERNEL_PATHS = new Set([
   '/providers/rules',
   '/connections',
   '/memory',
+  '/traffic',
+  '/logs',
 ])
 
 function json(body: unknown, status = 200): Response {
diff --git a/shared/format.test.ts b/shared/format.test.ts
new file mode 100644
index 0000000..eb75367
--- /dev/null
+++ b/shared/format.test.ts
@@ -0,0 +1,30 @@
+import { describe, expect, it } from 'bun:test'
+import { formatBytes } from './format.ts'
+
+describe('流量数字', () => {
+  it('零和不满 1KB 的按字节显示，不带小数', () => {
+    expect(formatBytes(0)).toBe('0 B')
+    expect(formatBytes(512)).toBe('512 B')
+    expect(formatBytes(1023)).toBe('1023 B')
+  })
+
+  it('1024 的边界正好换档', () => {
+    expect(formatBytes(1024)).toBe('1 KB')
+    expect(formatBytes(1024 * 1024)).toBe('1 MB')
+  })
+
+  it('不满 100 保留一位小数，够 100 取整', () => {
+    expect(formatBytes(23.4 * 1024 * 1024)).toBe('23.4 MB')
+    expect(formatBytes(234.7 * 1024 * 1024)).toBe('235 MB')
+  })
+
+  it('超大数停在 TB 档', () => {
+    expect(formatBytes(5 * 1024 ** 4)).toBe('5 TB')
+    expect(formatBytes(5000 * 1024 ** 4)).toBe('5000 TB')
+  })
+
+  it('脏输入不炸', () => {
+    expect(formatBytes(-1)).toBe('0 B')
+    expect(formatBytes(Number.NaN)).toBe('0 B')
+  })
+})
diff --git a/shared/format.ts b/shared/format.ts
new file mode 100644
index 0000000..6be1949
--- /dev/null
+++ b/shared/format.ts
@@ -0,0 +1,20 @@
+/**
+ * 流量数字显示成人读的样子。连接页和仪表盘共用。
+ *
+ * 1024 进制。B 一档不带小数；其余不满 100 的保留一位小数，够 100 的取整——
+ * 「23.4 MB」有信息量，「234.7 MB」的小数就只是噪音了。
+ */
+const UNITS = ['B', 'KB', 'MB', 'GB', 'TB'] as const
+
+export function formatBytes(bytes: number): string {
+  if (!Number.isFinite(bytes) || bytes < 0) return '0 B'
+  let value = bytes
+  let unit = 0
+  while (value >= 1024 && unit < UNITS.length - 1) {
+    value /= 1024
+    unit += 1
+  }
+  const digits = unit === 0 || value >= 100 ? 0 : 1
+  // Number() 去掉无意义的尾零：1.0 KB → 1 KB
+  return `${String(Number(value.toFixed(digits)))} ${UNITS[unit]}`
+}
diff --git a/web/stream.test.ts b/web/stream.test.ts
new file mode 100644
index 0000000..aeb17b4
--- /dev/null
+++ b/web/stream.test.ts
@@ -0,0 +1,55 @@
+import { describe, expect, it } from 'bun:test'
+import { readLines } from './stream.ts'
+
+function streamOf(chunks: string[]): ReadableStream<Uint8Array> {
+  const encoder = new TextEncoder()
+  return new ReadableStream({
+    start(controller) {
+      for (const chunk of chunks) controller.enqueue(encoder.encode(chunk))
+      controller.close()
+    },
+  })
+}
+
+async function collect(chunks: string[]): Promise<string[]> {
+  const lines: string[] = []
+  for await (const line of readLines(streamOf(chunks))) lines.push(line)
+  return lines
+}
+
+describe('按行读流', () => {
+  it('一块里有好几行就拆开', async () => {
+    expect(await collect(['a\nb\nc\n'])).toEqual(['a', 'b', 'c'])
+  })
+
+  it('半行留在缓冲里，等下一块补齐', async () => {
+    // 分块边界和行边界无关，这是这个模块存在的全部理由
+    expect(await collect(['{"up":1,"do', 'wn":2}\n{"up":3}\n'])).toEqual([
+      '{"up":1,"down":2}',
+      '{"up":3}',
+    ])
+  })
+
+  it('流结束时缓冲里剩的半行也交出去', async () => {
+    expect(await collect(['a\nb'])).toEqual(['a', 'b'])
+  })
+
+  it('空行丢弃，\\r\\n 换行的 \\r 去掉', async () => {
+    expect(await collect(['a\r\n\n\nb\n'])).toEqual(['a', 'b'])
+  })
+
+  it('多字节字符跨块也不会读坏', async () => {
+    const bytes = new TextEncoder().encode('香港01\n')
+    const stream = new ReadableStream<Uint8Array>({
+      start(controller) {
+        // 故意从「港」字中间切开
+        controller.enqueue(bytes.slice(0, 4))
+        controller.enqueue(bytes.slice(4))
+        controller.close()
+      },
+    })
+    const lines: string[] = []
+    for await (const line of readLines(stream)) lines.push(line)
+    expect(lines).toEqual(['香港01'])
+  })
+})
diff --git a/web/stream.ts b/web/stream.ts
new file mode 100644
index 0000000..77a6913
--- /dev/null
+++ b/web/stream.ts
@@ -0,0 +1,33 @@
+/**
+ * 把持续吐数据的 HTTP 响应体读成一行一行。
+ *
+ * 内核的 /traffic、/memory、/logs 在没有 WebSocket 升级时退化成分块流，一行一个
+ * JSON。分块的边界和行的边界无关：一次 read 可能拿到半行，也可能拿到好几行。
+ * 这里把它们缝成完整的行再交出去，空行丢弃。
+ *
+ * 停止读取的方式是中止 fetch（AbortController）或 break 出循环，两种都会走到
+ * finally 里把流取消掉，不留悬着的连接。
+ */
+export async function* readLines(body: ReadableStream<Uint8Array>): AsyncGenerator<string> {
+  const reader = body.getReader()
+  const decoder = new TextDecoder()
+  let buffer = ''
+  try {
+    for (;;) {
+      const { done, value } = await reader.read()
+      if (done) break
+      buffer += decoder.decode(value, { stream: true })
+      const parts = buffer.split('\n')
+      // 最后一段可能是半行，留在缓冲里等下一块补齐
+      buffer = parts.pop() ?? ''
+      for (const part of parts) {
+        const line = part.endsWith('\r') ? part.slice(0, -1) : part
+        if (line) yield line
+      }
+    }
+    const rest = buffer.trim()
+    if (rest) yield rest
+  } finally {
+    await reader.cancel().catch(() => {})
+  }
+}