caigee-cmd · caigee-cmd · Sep 4, 2026 · Sep 4, 2026 · Sep 4, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,8 +7,16 @@ Write each change in both `### English` and `### 中文` under `## Unreleased`.
 
 ### English
 
+- Distinguish hard quota exhaustion from prompt token limits, and cool exhausted accounts until the next local midnight
+- Keep explicitly identified conversations on one account with bounded in-memory session affinity, same-provider/region escape, and routing-source request logs
+- Add stateless Anthropic Messages and OpenAI Responses adapters for text and function-tool conversations
+
 ### 中文
 
+- 区分账号额度耗尽和请求 token 超限，真正额度耗尽的账号冷却到本地次日凌晨
+- 增加基于显式会话标识的有界内存会话粘性路由、同提供方同区域逃逸，以及路由来源请求日志
+- 新增面向文本与函数工具对话的无状态 Anthropic Messages 与 OpenAI Responses 适配层
+
 ## 0.2.41 - 2026-09-03
 
 ### English

diff --git a/README.md b/README.md
@@ -21,7 +21,7 @@
 
 ## 功能
 
-- **OpenAI 兼容代理**：`/v1/chat/completions`、`/v1/models`，流式/非流式、工具调用、`reasoning_content`
+- **OpenAI / Anthropic 兼容代理**：`/v1/chat/completions`、`/v1/responses`、`/v1/messages`、`/v1/models`；支持流式/非流式、文本、图片与函数工具调用；文件输入会明确拒绝。`messages` / `responses` 当前为无状态适配层，不支持服务端会话或上游专属工具。
 - **多渠道账号池**：Qoder 国际版 / 国内版、WorkBuddy 国际版 / 国内版、Trae 国内版 Solo；地域隔离、账号固定、并发限制、冷却与同族故障切换
 - **常驻热 worker**：每账号独立 Node 进程与运行目录，鉴权、WASM 编码和云端 SSE 连接保持热；预热后小对话延迟约 1-2 秒，按请求拉起 CLI 的方案通常要 10 秒以上
 - **多种登录方式**：浏览器 Device Flow OAuth、PAT、`qoder-native-v1` 凭证导入/导出
@@ -50,7 +50,7 @@ Base URL: http://127.0.0.1:3010/v1
 API Key:  <首次启动时生成的 Key>
 ```
 
-不指定账号时，调度器自动选择可用账号；需要固定账号时加请求头 `X-Qoder-Account: acc_...`。curl / PowerShell 示例见 [部署说明](deploy/README.md)。
+不指定账号时，调度器自动选择可用账号；需要固定账号时加请求头 `X-Qoder-Account: acc_...`。除 Chat Completions 外，也可使用 Anthropic `POST /v1/messages` 与 OpenAI `POST /v1/responses`；两者要求请求携带完整对话，不支持 `previous_response_id` / `conversation` 服务端续接。curl / PowerShell 示例见 [部署说明](deploy/README.md)。
 
 ## 工作方式
 
@@ -82,10 +82,9 @@ CLI2API 是本地网关：不提供账号、额度或官方 API 服务，不做
 **进行中**
 
 - Qoder 国内版与 WorkBuddy 的真账号验收（登录、故障切换、混合账号池）
-- Phase I：Anthropic `/v1/messages` 入口适配层与统一会话契约
-
 **已支持**
 
+- Anthropic `/v1/messages` 与 OpenAI `/v1/responses` 的无状态文本 / 函数工具适配层
 - WorkBuddy 每日签到与 token 保活（账号级开关，默认关闭；控制台可立即签到 / 刷新积分）
 
 **规划中**

diff --git a/README_EN.md b/README_EN.md
@@ -21,7 +21,7 @@ Long-lived workers, multi-account scheduling, one-command Docker start.
 
 ## Features
 
-- **OpenAI-compatible proxy**: `/v1/chat/completions`, `/v1/models` — streaming/non-streaming, tool calls, `reasoning_content`
+- **OpenAI / Anthropic-compatible proxy**: `/v1/chat/completions`, `/v1/responses`, `/v1/messages`, `/v1/models` — streaming/non-streaming text, images, and function tools; file inputs are rejected explicitly. `messages` / `responses` are stateless adapters today and do not support server-side conversations or upstream-specific tools.
 - **Multi-channel account pool**: Qoder Global / Qoder CN, WorkBuddy Global / WorkBuddy CN, Trae CN Solo — region isolation, account pinning, concurrency limits, cooldowns, and same-family failover
 - **Long-lived warm workers**: one isolated Node process and runtime directory per account keeps authentication, WASM encoding, and cloud SSE connections warm. Typical small-chat latency is ~1-2s after warmup, versus ~10s+ for spawn-per-request wrappers
 - **Multiple login methods**: browser Device Flow OAuth, PAT, and `qoder-native-v1` credential import/export
@@ -50,7 +50,7 @@ Base URL: http://127.0.0.1:3010/v1
 API Key:  <the key printed on first startup>
 ```
 
-Without an account header the scheduler picks a ready account; pin a request with the `X-Qoder-Account: acc_...` header. curl / PowerShell examples in the [deployment guide](deploy/README.md).
+Without an account header the scheduler picks a ready account; pin a request with the `X-Qoder-Account: acc_...` header. Anthropic `POST /v1/messages` and OpenAI `POST /v1/responses` are also available; both require the complete conversation in each request and do not support server-side continuation through `previous_response_id` / `conversation`. curl / PowerShell examples in the [deployment guide](deploy/README.md).
 
 ## How it works
 
@@ -82,10 +82,9 @@ CLI2API is a local gateway: it does not provide accounts, quotas, or an official
 **In progress**
 
 - Live-account acceptance for Qoder CN and WorkBuddy (login, failover, mixed account pools)
-- Phase I: Anthropic `/v1/messages` ingress adapter and the canonical conversation contract
-
 **Supported**
 
+- Stateless text, image, and function-tool adapters for Anthropic `/v1/messages` and OpenAI `/v1/responses`
 - WorkBuddy daily check-in and token keepalive (per-account opt-in, off by default; console can check in now / refresh credits)
 
 **Planned**

diff --git a/frontend/src/api/logs.ts b/frontend/src/api/logs.ts
@@ -27,6 +27,7 @@ export type RequestLog = {
   mapped_model?: string
   account_id?: string
   provider?: string
+  routing?: string
   prompt_tokens?: number | null
   completion_tokens?: number | null
   cache_read_tokens?: number | null

diff --git a/frontend/src/components/account/AccountCard.tsx b/frontend/src/components/account/AccountCard.tsx
@@ -17,7 +17,7 @@ import {
 import gsap from 'gsap'
 import { ProviderMark } from '@/components/ProviderMark'
 import { CompactSwitch } from '@/components/ui/CompactSwitch'
-import { AccountCardSkeleton, SkeletonBlock } from '@/components/ui/PageSkeletons'
+import { AccountCardSkeleton } from '@/components/ui/PageSkeletons'
 import { QuotaMeter } from '@/components/account/QuotaMeter'
 import { RuntimeMeter } from '@/components/account/RuntimeMeter'
 import {
@@ -35,7 +35,6 @@ type Translate = (key: string, vars?: Record<string, string | number>) => string
 type Props = {
   account: AccountRow
   busyKind: AccountBusyKind | ''
-  detailsLoading?: boolean
   authPanelOpen: boolean
   authUrl?: string
   note?: string
@@ -82,7 +81,6 @@ function checkinStatusFor(account: AccountRow, t: Translate) {
 export function AccountCard({
   account,
   busyKind,
-  detailsLoading = false,
   authPanelOpen,
   authUrl,
   note,
@@ -133,25 +131,6 @@ export function AccountCard({
   const provider = accountProviderLabel(account.provider, account.region, t)
   const checkin = checkinStatusFor(account, t)
 
-  const detailsSkeleton = (
-    <>
-      <div className="rounded-2xl border border-border bg-surface-secondary/45 p-2.5">
-        <SkeletonBlock className="h-3 w-20" />
-        <SkeletonBlock className="mt-2.5 h-2 w-full" />
-        <div className="mt-2.5 grid grid-cols-3 gap-2 border-t border-separator pt-2.5">
-          <SkeletonBlock className="h-7 w-full" />
-          <SkeletonBlock className="h-7 w-full" />
-          <SkeletonBlock className="h-7 w-full" />
-        </div>
-      </div>
-      <div className="min-h-[58px] rounded-2xl border border-border bg-surface-secondary/25 p-2.5">
-        <SkeletonBlock className="h-3 w-20" />
-        <SkeletonBlock className="mt-2 h-1.5 w-full rounded-[1px]" />
-        <SkeletonBlock className="mt-2 h-3 w-36" />
-      </div>
-    </>
-  )
-
   useLayoutEffect(() => {
     const chip = chipRef.current
     if (!chip) return
@@ -201,7 +180,7 @@ export function AccountCard({
   // status are never left on screen while the account is re-probed.
   if (busyKind === 'refresh') {
     return (
-      <div data-gsap-reveal aria-busy="true" aria-label={t('refreshingAccount')}>
+      <div data-gsap-reveal className="min-h-[280px]" aria-busy="true" aria-label={t('refreshingAccount')}>
         <AccountCardSkeleton />
       </div>
     )
@@ -211,8 +190,7 @@ export function AccountCard({
     <Card
       data-gsap-reveal
       data-state={state}
-      aria-busy={detailsLoading}
-      className="account-card overflow-hidden p-0"
+      className="account-card min-h-[280px] overflow-hidden p-0"
     >
       <Card.Header className="flex-row items-start justify-between gap-2.5 px-3 pt-2.5 pb-1.5">
         <div className="flex min-w-0 items-center gap-2.5">
@@ -230,31 +208,50 @@ export function AccountCard({
           </div>
         </div>
         <div className="flex max-w-[48%] shrink-0 flex-wrap justify-end gap-1.5">
-          {detailsLoading ? <SkeletonBlock className="h-5 w-16" /> : (
-            <>
-              <span ref={chipRef}>
-                <Chip size="sm" variant="soft" color={stateColor}>{stateCopy}</Chip>
-              </span>
-              {modelCooldowns.length ? <Chip size="sm" variant="soft" color="warning">{t('partialCooling')}</Chip> : null}
-            </>
-          )}
+          <span ref={chipRef}>
+            {modelCooldowns.length && state === 'cooling' ? (
+              <Tooltip>
+                <Tooltip.Trigger>
+                  <span className="inline-flex cursor-help"><Chip size="sm" variant="soft" color={stateColor}>{stateCopy}</Chip></span>
+                </Tooltip.Trigger>
+                <Tooltip.Content>
+                  <div className="space-y-1.5">
+                    <div className="font-medium">{t('partialCooling')}</div>
+                    {modelCooldowns.map(([model, until]) => (
+                      <div key={model} className="flex items-center justify-between gap-4 text-xs">
+                        <span>{model}</span>
+                        <span className="mono text-foreground/65">{cooldownLabel(until)}</span>
+                      </div>
+                    ))}
+                  </div>
+                </Tooltip.Content>
+              </Tooltip>
+            ) : (
+              <Chip size="sm" variant="soft" color={stateColor}>{stateCopy}</Chip>
+            )}
+          </span>
+          {modelCooldowns.length && state !== 'cooling' ? (
+            <Tooltip>
+              <Tooltip.Trigger>
+                <span className="inline-flex cursor-help"><Chip size="sm" variant="soft" color="warning">{t('partialCooling')}</Chip></span>
+              </Tooltip.Trigger>
+              <Tooltip.Content>
+                <div className="space-y-1.5">
+                  <div className="font-medium">{t('partialCooling')}</div>
+                  {modelCooldowns.map(([model, until]) => (
+                    <div key={model} className="flex items-center justify-between gap-4 text-xs">
+                      <span>{model}</span>
+                      <span className="mono text-foreground/65">{cooldownLabel(until)}</span>
+                    </div>
+                  ))}
+                </div>
+              </Tooltip.Content>
+            </Tooltip>
+          ) : null}
         </div>
       </Card.Header>
 
       <Card.Content className="gap-2 px-3 pb-2">
-        {detailsLoading ? detailsSkeleton : <>
-        {modelCooldowns.length ? (
-          <div className="flex items-start gap-2 rounded-2xl border border-warning/25 bg-warning/5 px-2.5 py-2 text-[11px] text-warning-foreground dark:text-warning">
-            <span className="status-dot mt-0.5 shrink-0" data-state="warn" />
-            <div className="min-w-0">
-              <div className="font-medium">{t('partialCooling')}</div>
-              <div className="mt-0.5 break-words text-warning-foreground/80 dark:text-warning/80">
-                {modelCooldowns.slice(0, 3).map(([model, until]) => `${model} ${cooldownLabel(until)}`).join(' · ')}
-                {modelCooldowns.length > 3 ? ` · +${modelCooldowns.length - 3}` : ''}
-              </div>
-            </div>
-          </div>
-        ) : null}
         <div className="rounded-2xl border border-border bg-surface-secondary/45 p-2">
           <RuntimeMeter state={state} label={t('runtimeState')} stateCopy={stateCopy} />
           <div className="mt-2 grid grid-cols-3 gap-2 border-t border-separator pt-2 text-[10px] text-foreground/65">
@@ -326,7 +323,6 @@ export function AccountCard({
             </div>
           </div>
         ) : null}
-        </>}
       </Card.Content>
 
       {authPanelOpen && account.enabled ? (

diff --git a/frontend/src/components/ui/PageSkeletons.tsx b/frontend/src/components/ui/PageSkeletons.tsx
@@ -126,7 +126,7 @@ export function OverviewPageSkeleton() {
 
 export function AccountCardSkeleton() {
   return (
-    <article className="overflow-hidden rounded-3xl border border-border bg-surface">
+    <article className="min-h-[280px] overflow-hidden rounded-3xl border border-border bg-surface">
       <div className="space-y-2 px-3 pt-2.5 pb-1.5">
         <div className="flex items-start justify-between gap-2.5">
           <div className="flex min-w-0 items-center gap-2.5">
@@ -169,7 +169,7 @@ export function AccountCardSkeleton() {
   )
 }
 
-export function AccountsListSkeleton({ count = 6 }: { count?: number }) {
+export function AccountsListSkeleton({ count = 5 }: { count?: number }) {
   return (
     <section className="grid gap-2.5 lg:grid-cols-2 xl:grid-cols-3 2xl:grid-cols-4" aria-label="Loading accounts">
       {Array.from({ length: count }, (_, index) => <AccountCardSkeleton key={index} />)}

diff --git a/frontend/src/i18n/messages.ts b/frontend/src/i18n/messages.ts
@@ -182,6 +182,11 @@ export const messages: Record<Lang, Dict> = {
     logsLevelWarn: 'Warn',
     logsLevelError: 'Error',
     logsAttempts: 'Attempts',
+    logsRouting: 'Routing',
+    logsRouting_pool: 'Pool',
+    logsRouting_pin: 'Pinned',
+    logsRouting_sticky: 'Session sticky',
+    logsRouting_sticky_escape: 'Session escape',
     logsNoAttempts: 'No attempt details.',
     logsDetailTitle: 'Request detail',
     logsShownTotal: '{shown} / {total}',
@@ -709,6 +714,11 @@ export const messages: Record<Lang, Dict> = {
     logsLevelWarn: '警告',
     logsLevelError: '错误',
     logsAttempts: '尝试记录',
+    logsRouting: '路由方式',
+    logsRouting_pool: '常规池',
+    logsRouting_pin: '显式指定',
+    logsRouting_sticky: '会话粘性',
+    logsRouting_sticky_escape: '粘性逃逸',
     logsNoAttempts: '没有尝试详情。',
     logsDetailTitle: '请求详情',
     logsShownTotal: '{shown} / {total}',

diff --git a/frontend/src/pages/AccountsPage.tsx b/frontend/src/pages/AccountsPage.tsx
@@ -47,11 +47,8 @@ export function AccountsPage() {
   const { t } = useI18n()
   const [accounts, setAccounts] = useState<AccountRow[]>([])
   const [accountsLoading, setAccountsLoading] = useState(true)
-  const [accountsRefreshing, setAccountsRefreshing] = useState(false)
-  const [refreshingById, setRefreshingById] = useState<Record<string, boolean>>({})
   const rows = accounts
   const hasAccounts = rows.length > 0
-  const refreshing = accountsRefreshing && hasAccounts
   const [addOpen, setAddOpen] = useState(false)
   const [busy, setBusy] = useState<AccountBusy | null>(null)
   const [patById, setPatById] = useState<Record<string, string>>({})
@@ -75,43 +72,29 @@ export function AccountsPage() {
   const [priorityById, setPriorityById] = useState<Record<string, number>>({})
   const [quotaRefreshing, setQuotaRefreshing] = useState(false)
   const reloadAccounts = useCallback(async (refresh = false) => {
-    if (refresh) setAccountsRefreshing(true)
     try {
       const response = await fetchAccounts(refresh)
       const next = response.data || EMPTY_ACCOUNTS
       setAccounts(next)
       return next
     } finally {
       setAccountsLoading(false)
-      if (refresh) setAccountsRefreshing(false)
     }
   }, [])
 
   const refreshAccountsInBatches = useCallback(async (ids: string[], forceQuota = true) => {
     const accountIds = [...new Set(ids)]
     if (!accountIds.length) return
-    setAccountsRefreshing(true)
-    try {
-      for (let start = 0; start < accountIds.length; start += ACCOUNT_REFRESH_BATCH_SIZE) {
-        const batch = accountIds.slice(start, start + ACCOUNT_REFRESH_BATCH_SIZE)
-        await Promise.all(batch.map(async (id) => {
-          setRefreshingById((current) => ({ ...current, [id]: true }))
-          try {
-            const refreshed = await refreshAccount(id, { quota: forceQuota })
-            setAccounts((current) => current.map((account) => account.id === id ? refreshed : account))
-          } catch (error) {
-            setNoteById((current) => ({ ...current, [id]: error instanceof Error ? error.message : String(error) }))
-          } finally {
-            setRefreshingById((current) => {
-              const next = { ...current }
-              delete next[id]
-              return next
-            })
-          }
-        }))
-      }
-    } finally {
-      setAccountsRefreshing(false)
+    for (let start = 0; start < accountIds.length; start += ACCOUNT_REFRESH_BATCH_SIZE) {
+      const batch = accountIds.slice(start, start + ACCOUNT_REFRESH_BATCH_SIZE)
+      await Promise.all(batch.map(async (id) => {
+        try {
+          const refreshed = await refreshAccount(id, { quota: forceQuota })
+          setAccounts((current) => current.map((account) => account.id === id ? refreshed : account))
+        } catch (error) {
+          setNoteById((current) => ({ ...current, [id]: error instanceof Error ? error.message : String(error) }))
+        }
+      }))
     }
   }, [])
 
@@ -400,7 +383,6 @@ export function AccountsPage() {
           ))}
         </div>
         <div className="flex flex-wrap items-center gap-2">
-          {accountsRefreshing ? <span className="text-xs text-muted" role="status">{t('refreshing')}</span> : null}
           <Button size="sm" variant="secondary" isPending={quotaRefreshing} onPress={() => void onRefreshCredits()}>
             {t('refreshCredits')}
           </Button>
@@ -472,7 +454,7 @@ export function AccountsPage() {
         </section>
       ) : null}
 
-      {!refreshing && !hasAccounts ? (
+      {!hasAccounts ? (
         <EmptyPanel
           className="rounded-3xl border border-dashed border-border"
           icon={<BrandMark size={28} />}
@@ -491,13 +473,12 @@ export function AccountsPage() {
         />
       ) : null}
 
-      <section className="grid gap-2.5 lg:grid-cols-2 xl:grid-cols-3 2xl:grid-cols-4" aria-busy={refreshing}>
+      <section className="grid auto-rows-[minmax(280px,auto)] gap-2.5 lg:grid-cols-2 xl:grid-cols-3 2xl:grid-cols-4">
         {pagedRows.map((account) => (
           <AccountCard
             key={account.id}
             account={account}
             busyKind={busy?.id === account.id ? busy.kind : ''}
-            detailsLoading={Boolean(refreshingById[account.id])}
             authPanelOpen={authPanelId === account.id}
             authUrl={urlById[account.id]}
             note={noteById[account.id]}

diff --git a/frontend/src/pages/LogsPage.tsx b/frontend/src/pages/LogsPage.tsx
@@ -172,6 +172,15 @@ export function LogsPage() {
     return accountProviderLabel(provider, account?.region, t)
   }
 
+  function routingLabel(routing?: string) {
+    switch (routing) {
+      case 'pin': return t('logsRouting_pin')
+      case 'sticky': return t('logsRouting_sticky')
+      case 'sticky_escape': return t('logsRouting_sticky_escape')
+      default: return t('logsRouting_pool')
+    }
+  }
+
   const accountOptions = useMemo(
     () => (accounts || []).map((account) => ({ id: account.id, label: account.name || account.id })),
     [accounts],
@@ -785,6 +794,7 @@ export function LogsPage() {
                     [t('logsColLatency'), formatLatency(selected?.latency_ms)],
                     [t('logsColTTFT'), formatLatency(selected?.ttfb_ms)],
                     [t('logsColStream'), selected?.stream ? t('logsStreamYes') : t('logsStreamNo')],
+                    [t('logsRouting'), selected ? routingLabel(selected.routing) : '—'],
                   ].map(([label, value]) => (
                     <div key={String(label)}>
                       <dt className="text-[11px] text-muted">{label}</dt>