feat(quota): report A6API credit usage

docs-site/src/content/docs/guides/providers.md

@@ -308,6 +308,31 @@ Provider plan, and CLI auth bridging for Go/Pro subscriptions is not yet availab
 > hosts and schemas and are not routed by this preset.
 > Live discovery for this preset is capped at a 1 MiB response and 256 raw model rows.
 
+### A6API credit quota
+
+A custom `openai-chat` provider using `authMode: "key"` and the canonical
+`https://api.a6api.com` or `https://api.a6api.com/v1` base URL receives an A6API credit meter in
+the dashboard and from `ocx account refresh <provider>`. The provider name is arbitrary; detection
+uses the canonical HTTPS endpoint. The meter converts A6API token units into USD using the account's
+hard credit limit and displays the percentage consumed plus remaining credit. Token expiration is
+not shown as a quota reset because expiration does not imply that credit replenishes.
+
+```json
+{
+  "providers": {
+    "my-a6": {
+      "adapter": "openai-chat",
+      "authMode": "key",
+      "baseUrl": "https://api.a6api.com/v1",
+      "apiKey": "${A6API_API_KEY}"
+    }
+  }
+}
+```
+
+Quota probes send only the active key to the canonical A6API host and reject redirects. Malformed,
+negative, or internally inconsistent billing totals produce no report rather than a misleading bar.
+
 > **Tencent Cloud Coding Plan usage restriction:** Tencent documents this subscription for
 > interactive coding tools only. General API automation, custom application backends, and
 > non-interactive batch use are prohibited and may cause the plan key to be suspended.

docs-site/src/content/docs/ja/guides/providers.md

@@ -230,6 +230,15 @@ API アクセスには Provider プランが必要で、Go/Pro サブスクリ
 > エンドポイントはホストとスキーマが異なるため、このプリセットではルーティングされません。
 > このプリセットのライブディスカバリーは、レスポンス 1 MiB、モデルの生行 256 件が上限です。
 
+### A6API クレジットクォータ
+
+`openai-chat`、`authMode: "key"`、正規の `https://api.a6api.com` または
+`https://api.a6api.com/v1` を使うカスタムプロバイダーでは、ダッシュボードと
+`ocx account refresh <provider>` に A6API クレジット使用量が表示されます。プロバイダー名は任意です。
+アカウントの hard credit limit を基準にトークン単位を USD に換算し、使用率と残高を表示します。トークン期限は補充を意味しないため、クォータの
+リセットとしては表示しません。アクティブキーだけを正規ホストへ送信し、リダイレクトを拒否します。負数や
+整合しない請求合計からはレポートを生成しません。
+
 > **Tencent Cloud Coding Plan の利用制限:** Tencent はこのサブスクリプションを対話型
 > コーディングツール専用としています。一般的な API 自動化、カスタムアプリのバックエンド、
 > 非対話型バッチ利用は禁止されており、プランキーが停止される場合があります。

docs-site/src/content/docs/ko/guides/providers.md

@@ -229,6 +229,15 @@ Bearer API 키를 사용합니다. registry가 소유하는 DeepInfra 모델 목
 > 호스트와 스키마가 다르므로 이 프리셋으로 라우팅되지 않습니다.
 > 이 프리셋의 실시간 검색은 응답 1 MiB와 원시 모델 행 256개로 제한됩니다.
 
+### A6API 크레딧 쿼터
+
+`openai-chat`, `authMode: "key"`, 공식 `https://api.a6api.com` 또는
+`https://api.a6api.com/v1` 주소를 사용하는 사용자 지정 프로바이더는 대시보드와
+`ocx account refresh <provider>`에서 A6API 크레딧 사용량을 표시합니다. 프로바이더 이름은 자유롭게 정할 수
+있습니다. 계정의 hard credit limit을 기준으로 토큰 단위를 USD로 환산해 사용률과 남은 크레딧을 표시하며, 토큰 만료는 충전을 뜻하지 않으므로 쿼터
+리셋으로 표시하지 않습니다. 활성 키만 공식 호스트로 전송하고 리디렉션을 거부하며, 음수이거나 서로 일치하지
+않는 결제 합계에는 보고서를 만들지 않습니다.
+
 > **Tencent Cloud Coding Plan 사용 제한:** Tencent는 이 구독을 대화형 코딩 도구 전용으로
 > 안내합니다. 일반 API 자동화, 사용자 애플리케이션 백엔드 및 비대화형 일괄 호출은 금지되며
 > 플랜 키가 정지될 수 있습니다.

docs-site/src/content/docs/ru/guides/providers.md

@@ -241,6 +241,16 @@ endpoint в него не входят. Ключи создаются в [Hyperb
 > схемы и этим пресетом не маршрутизируются.
 > Для этого пресета live discovery ограничен ответом размером 1 MiB и 256 исходными строками моделей.
 
+### Квота кредитов A6API
+
+Пользовательский провайдер с `openai-chat`, `authMode: "key"` и каноническим адресом
+`https://api.a6api.com` или `https://api.a6api.com/v1` показывает расход кредитов A6API в
+дашборде и в `ocx account refresh <provider>`. Имя провайдера может быть любым. Единицы токенов
+пересчитываются в USD по hard credit limit учётной записи; отображаются процент расхода и остаток. Срок действия токена не считается
+сбросом квоты, поскольку он не означает пополнение. Только активный ключ отправляется на
+канонический хост, перенаправления отклоняются, а отрицательные или несогласованные итоги биллинга
+не создают отчёт.
+
 > **Ограничение Tencent Cloud Coding Plan:** Tencent разрешает использовать эту подписку только
 > в интерактивных инструментах программирования. Автоматизация общего API, серверы пользовательских
 > приложений и неинтерактивные пакетные вызовы запрещены и могут привести к блокировке ключа плана.

docs-site/src/content/docs/zh-cn/guides/providers.md

@@ -213,6 +213,14 @@ bearer 密钥；API 访问需要 Provider 套餐，Go/Pro 订阅用户的 CLI 
 > 专用 Truss `predict` 端点使用不同的主机和请求 schema，不由此预设路由。
 > 该预设的实时发现上限为 1 MiB 响应和 256 条原始模型记录。
 
+### A6API 信用额度
+
+使用 `openai-chat`、`authMode: "key"` 以及规范地址 `https://api.a6api.com` 或
+`https://api.a6api.com/v1` 的自定义提供商，会在仪表板和 `ocx account refresh <provider>`
+中显示 A6API 信用使用情况；提供商名称可以自定义。系统依据账户的 hard credit limit 将令牌单位换算为 USD，并显示已用百分比和剩余额度。
+令牌到期不代表额度补充，因此不会显示为配额重置。只有当前活动密钥会发送到规范主机，重定向会被拒绝；负数
+或内部不一致的计费总数不会生成报告。
+
 > **腾讯云 Coding Plan 使用限制：**腾讯将此订阅限定为交互式编程工具使用。禁止通用 API
 > 自动化、自定义应用后端和非交互式批量调用；违规使用可能导致套餐密钥被停用。
 

src/providers/quota.ts

@@ -10,6 +10,7 @@ import { resolveEnvValue } from "../config";
 import { getValidAccessToken, getValidAccessTokenForAccount } from "../oauth";
 import { getAccountCredential, getAccountSet, getCredential } from "../oauth/store";
 import { antigravityUserAgent } from "../adapters/client-fingerprint";
+import { apiKeyPoolEntryId } from "./api-keys";
 import { getProviderRegistryEntry, providerCodexAccountMode } from "./registry";
 import type { OcxConfig, OcxProviderConfig } from "../types";
 import { isCanonicalOpenAiForwardProvider, OPENAI_CODEX_PROVIDER_ID } from "./openai-tiers";
@@ -32,6 +33,7 @@ const CACHE_TTL_MS = 5 * 60_000;
 const REQUEST_TIMEOUT_MS = 8_000;
 const KIMI_CODE_BASE_URL = "https://api.kimi.com/coding/v1";
 const KIMI_CODE_USAGE_URL = `${KIMI_CODE_BASE_URL}/usages`;
+const A6API_BASE_URL = "https://api.a6api.com";
 /** Keep a failed probe's previous row at most this long before dropping it. */
 const LAST_GOOD_MAX_AGE_MS = CODEX_CAPACITY_MAX_QUOTA_AGE_MS;
 const nativeMainReportGenerations = new WeakMap<ProviderQuotaReport, number>();
@@ -43,6 +45,8 @@ export function setProviderQuotaBeforePublishForTests(
 ): void {
   providerQuotaBeforePublishForTests = hook;
 }
+const TERMINAL_QUOTA_FAILURE = Symbol("terminal-quota-failure");
+type ProviderQuotaProbeResult = ProviderQuotaReport | null | typeof TERMINAL_QUOTA_FAILURE;
 
 export interface ProviderQuotaWindow {
   label: string;
@@ -89,7 +93,13 @@ export function clearProviderQuotaCache(): void {
 
 function cacheKey(config: OcxConfig): string {
   const providers = Object.entries(config.providers)
-    .map(([name, provider]) => `${name}:${provider.adapter}:${provider.authMode ?? "key"}:${providerCodexAccountMode(name, provider) ?? "none"}:${provider.disabled === true ? "off" : "on"}:${provider.baseUrl}`)
+    .map(([name, provider]) => {
+      const resolvedKey = typeof provider.apiKey === "string"
+        ? resolveEnvValue(provider.apiKey)?.trim()
+        : undefined;
+      const activeKeyId = resolvedKey ? apiKeyPoolEntryId(resolvedKey) : "none";
+      return `${name}:${provider.adapter}:${provider.authMode ?? "key"}:${providerCodexAccountMode(name, provider) ?? "none"}:${provider.disabled === true ? "off" : "on"}:${provider.baseUrl}:${activeKeyId}`;
+    })
     .sort()
     .join("|");
   return `${config.defaultProvider}|${providers}`;
@@ -228,6 +238,77 @@ function isBuiltInChatGptForwardProvider(name: string, provider: OcxProviderConf
   return name === OPENAI_CODEX_PROVIDER_ID && isCanonicalOpenAiForwardProvider(provider);
 }
 
+function isCanonicalA6apiBaseUrl(baseUrl: string): boolean {
+  const normalized = normalizedBaseUrl(baseUrl);
+  return normalized === A6API_BASE_URL || normalized === `${A6API_BASE_URL}/v1`;
+}
+
+function a6apiPayload(value: unknown): Record<string, unknown> | null {
+  const body = asRecord(value);
+  return asRecord(body?.data) ?? body;
+}
+
+function firstFinite(record: Record<string, unknown> | null, names: string[]): number | undefined {
+  if (!record) return undefined;
+  for (const name of names) {
+    const value = toFiniteNumber(record[name]);
+    if (value !== undefined) return value;
+  }
+  return undefined;
+}
+
+async function fetchA6apiQuota(provider: string, config: OcxProviderConfig): Promise<ProviderQuotaProbeResult> {
+  // Never send a configured API key to a lookalike host or through a redirect.
+  if (!isCanonicalA6apiBaseUrl(config.baseUrl)) return null;
+  const apiKey = resolveEnvValue(config.apiKey)?.trim();
+  if (!apiKey) return null;
+  const headers = { Accept: "application/json", Authorization: `Bearer ${apiKey}` } as const;
+  const [subscriptionResponse, tokenResponse] = await Promise.all([
+    fetch(`${A6API_BASE_URL}/dashboard/billing/subscription`, {
+      headers, redirect: "error", signal: AbortSignal.timeout(REQUEST_TIMEOUT_MS),
+    }),
+    fetch(`${A6API_BASE_URL}/api/usage/token/`, {
+      headers, redirect: "error", signal: AbortSignal.timeout(REQUEST_TIMEOUT_MS),
+    }),
+  ]);
+  if (!subscriptionResponse.ok || !tokenResponse.ok) {
+    const statuses = [subscriptionResponse.status, tokenResponse.status];
+    // 408/429 are transient (timeout/throttle), not invalid-account signals: keep the
+    // last-good row like 5xx/network failures. 401/403 (bad key) and 404 (contract change)
+    // stay terminal.
+    return statuses.some(status => status >= 400 && status < 500 && status !== 429 && status !== 408)
+      ? TERMINAL_QUOTA_FAILURE
+      : null;
+  }
+  const subscription = a6apiPayload(await subscriptionResponse.json().catch(() => null));
+  const token = a6apiPayload(await tokenResponse.json().catch(() => null));
+  const limitUsd = firstFinite(subscription, ["hard_limit_usd"]);
+  const grantedUnits = firstFinite(token, ["total_granted"]);
+  const usedUnits = firstFinite(token, ["total_used"]);
+  const availableUnits = firstFinite(token, ["total_available"]);
+  const reconciledUnits = usedUnits !== undefined && availableUnits !== undefined
+    ? usedUnits + availableUnits
+    : undefined;
+  const reconciliationTolerance = grantedUnits !== undefined
+    ? Math.abs(grantedUnits) * 1e-9
+    : 0;
+  if (limitUsd === undefined || grantedUnits === undefined || usedUnits === undefined
+    || availableUnits === undefined || limitUsd <= 0 || grantedUnits <= 0
+    || usedUnits < 0 || availableUnits < 0
+    || reconciledUnits === undefined
+    || Math.abs(reconciledUnits - grantedUnits) > reconciliationTolerance) return TERMINAL_QUOTA_FAILURE;
+  const usdPerUnit = limitUsd / grantedUnits;
+  const usedUsd = usedUnits * usdPerUnit;
+  const remainingUsd = Math.max(0, availableUnits * usdPerUnit);
+  const percent = normalizePercent((usedUsd / limitUsd) * 100);
+  if (percent === undefined) return TERMINAL_QUOTA_FAILURE;
+  const label = `API credits ($${remainingUsd.toFixed(2)} of $${limitUsd.toFixed(2)} remaining)`;
+  return report(provider, "a6api:billing", {
+    customWindows: [{ label, percent }],
+    updatedAt: Date.now(),
+  });
+}
+
 function report(
   provider: string,
   source: string,
@@ -671,7 +752,7 @@ export async function fetchProviderAccountQuotas(
 function normalizedBaseUrl(value: string): string | null {
   try {
     const url = new URL(value);
-    if (url.search || url.hash) return null;
+    if (url.username || url.password || url.search || url.hash) return null;
     return `${url.origin.toLowerCase()}${url.pathname.replace(/\/+$/, "")}`;
   } catch {
     return null;
@@ -1079,7 +1160,7 @@ async function maybeFetchProviderQuota(
   config: OcxConfig,
   forceRefresh: boolean,
   prefetchedCodexSnapshot?: CodexAuthAccountsSnapshotPromise,
-): Promise<ProviderQuotaReport | null> {
+): Promise<ProviderQuotaProbeResult> {
   if (provider.disabled === true) return null;
   try {
     if (isBuiltInChatGptForwardProvider(name, provider)) {
@@ -1095,6 +1176,9 @@ async function maybeFetchProviderQuota(
     if (provider.authMode === "key" && isCanonicalKimiCodeBaseUrl(provider.baseUrl)) {
       return fetchKimiQuota(name, provider);
     }
+    if ((provider.authMode ?? "key") === "key" && isCanonicalA6apiBaseUrl(provider.baseUrl)) {
+      return fetchA6apiQuota(name, provider);
+    }
     return null;
   } catch {
     return null;
@@ -1125,20 +1209,24 @@ export async function fetchProviderQuotaReports(config: OcxConfig, forceRefresh
 
   const promise = (async (): Promise<ProviderQuotaResponse> => {
     const previous = cache && cache.key === key ? cache.response.reports : [];
-    const fresh = (await Promise.all(
+    const probeResults = await Promise.all(
       Object.entries(config.providers).map(([name, provider]) => (
-          maybeFetchProviderQuota(name, provider, config, forceRefresh, prefetchedCodexSnapshot)
-        )),
-      )).filter((item): item is ProviderQuotaReport => item !== null);
+        maybeFetchProviderQuota(name, provider, config, forceRefresh, prefetchedCodexSnapshot)
+      )),
+    );
+    const fresh = probeResults.filter((item): item is ProviderQuotaReport => item !== null && item !== TERMINAL_QUOTA_FAILURE);
+    const terminalFailures = new Set(
+      Object.keys(config.providers).filter((_, index) => probeResults[index] === TERMINAL_QUOTA_FAILURE),
+    );
     await providerQuotaBeforePublishForTests?.();
     let commitKey: string | null = null;
     if (epoch === invalidationEpoch) {
       const commitKeyCandidate = cacheKeyWithAggregationState(config);
       commitKey = typeof commitKeyCandidate === "string" ? commitKeyCandidate : await commitKeyCandidate;
     }
 
-    // Keep bounded last-good rows when a probe fails (e.g. transient upstream flake); never
-    // re-stamp their timestamps, and drop rows older than LAST_GOOD_MAX_AGE_MS.
+    // Keep bounded last-good rows when a probe fails transiently; terminal-invalid provider
+    // responses explicitly suppress their old row. Never re-stamp preserved timestamps.
     // Note: the cache key encodes the provider set (name/adapter/authMode/disabled/baseUrl),
     // so previous rows always correspond to currently configured, enabled providers — a
     // disabled or removed provider changes the key and starts from an empty previous set.
@@ -1159,6 +1247,11 @@ export async function fetchProviderQuotaReports(config: OcxConfig, forceRefresh
         generationMismatchedProviders.add(item.provider);
       }
     }
+    // Terminal-invalid probes suppress their previous row (transient failures keep it).
+    for (const provider of terminalFailures) {
+      byProvider.delete(provider);
+      generationMismatchedProviders.delete(provider);
+    }
 
     const response = { generatedAt: Date.now(), reports: [...byProvider.values()] };
     // Commit only when this probe still holds authority (no clear/force superseded it).