Merge pull request #1124 from kaitranntt/kai/feat/1123-self-pause-exhausted-accounts

feat(cliproxy): self-pause exhausted quota accounts

docs/project-roadmap.md

@@ -41,6 +41,7 @@ All major modularization work is complete. The codebase evolved from monolithic
 
 ### Recent Fixes
 
+- **2026-04-28**: **#1123** CLIProxy quota failover now uses the dashboard/manual pause mechanism for all quota-visible OAuth providers with CCS quota fetchers: Antigravity, Claude, Codex, Gemini CLI, and GitHub Copilot. When a healthy fallback exists, CCS moves the exhausted account token out of the live `auth/` folder into `auth-paused/`, marks the account paused for dashboard visibility, persists the cooldown for auto-resume, and still avoids self-pausing the last usable account.
 - **2026-04-28**: **#1115** CCS now exposes upstream CLIProxy session affinity as a first-class local managed setting. Users can inspect and toggle local `session-affinity` plus TTL from `ccs cliproxy routing affinity`, from the `/cliproxy` dashboard routing card, and through the local dashboard API. The generated local CLIProxy config now persists `routing.session-affinity` and `routing.session-affinity-ttl`, help/copy explains that CLIProxy prefers explicit session or thread identifiers before falling back to prompt-history hashing, and remote session-affinity management stays explicitly unsupported until upstream management APIs expose more than `routing.strategy`.
 - **2026-04-24**: **#1065** Local CLIProxy Plus is available again as an explicit opt-in backend through the community-maintained `kaitranntt/CLIProxyAPIPlus` fork. CCS keeps `original` as the default backend, no longer downgrades saved `backend: plus` configs to `original`, updates Plus release lookups to the maintained fork, and documents Plus as a targeted path for plus-only providers.
 - **2026-04-21**: CLIProxy quota failover now quarantines exhausted Claude and Antigravity accounts out of live rotation when a healthy fallback exists. CCS persists those quota-triggered pauses across launches, automatically resumes them after the configured cooldown window, and deliberately avoids auto-pausing the last available account so single-account setups still degrade gracefully instead of hard-locking themselves.

docs/system-architecture/provider-flows.md

@@ -261,7 +261,7 @@ async function checkRemoteProxyHealth(config: ResolvedProxyConfig): Promise<bool
 ### Overview
 
 Hybrid quota management enables automatic detection of exhausted accounts and failover to next available account.
-When CCS detects exhaustion and a healthy fallback exists, it also temporarily pauses the exhausted account out of CLIProxy rotation and automatically resumes that pause after the configured cooldown expires.
+When CCS detects exhaustion and a healthy fallback exists, it temporarily pauses the exhausted account out of CLIProxy rotation and automatically resumes that pause after the configured cooldown expires. This durable self-pause uses the same account registry and token movement path as dashboard/manual pause, so the dashboard shows the account as paused and CLIProxy cannot rediscover its token from the live `auth/` folder.
 
 ```
 +===========================================================================+
@@ -287,9 +287,11 @@ When CCS detects exhaustion and a healthy fallback exists, it also temporarily p
         |           +---> Check isPaused flag --> Skip if paused
         |           |
         |           +---> Fetch quota from provider API
-        |           |       - Gemini: /models endpoint
-        |           |       - Codex: /api/v1/account
-        |           |       - Kiro: /api/usage
+        |           |       - Antigravity: fetchAvailableModels
+        |           |       - Claude: policy limits endpoint
+        |           |       - Codex: ChatGPT usage windows
+        |           |       - Gemini CLI: Code Assist quota buckets
+        |           |       - GitHub Copilot: copilot_internal/user snapshots
         |           |
         |           +---> Detect tier (free/paid/unknown)
         |           |

src/cliproxy/account-safety.ts

@@ -690,6 +690,9 @@ export async function handleQuotaExhaustion(
   // Dynamic imports to avoid circular dependencies
   const { applyCooldown, findHealthyAccount } = await import('./quota-manager');
   const { setDefaultAccount, touchAccount } = await import('./account-manager');
+  const { loadOrCreateUnifiedConfig } = await import('../config/unified-config-loader');
+  const config = loadOrCreateUnifiedConfig();
+  const threshold = config.quota_management?.auto?.exhaustion_threshold ?? 5;
 
   // Apply cooldown to exhausted account
   applyCooldown(provider, accountId, cooldownMinutes);
@@ -698,7 +701,9 @@ export async function handleQuotaExhaustion(
   const alternative = await findHealthyAccount(provider, [accountId]);
 
   if (alternative) {
-    pauseAccountForQuotaCooldown(provider, accountId, cooldownMinutes);
+    if (alternative.lastQuota !== null && alternative.lastQuota >= threshold) {
+      pauseAccountForQuotaCooldown(provider, accountId, cooldownMinutes);
+    }
     setDefaultAccount(provider, alternative.id);
     touchAccount(provider, alternative.id);
     writeQuotaExhausted(accountId, alternative.id, cooldownMinutes);

src/cliproxy/executor/index.ts

@@ -106,6 +106,7 @@ import {
   handleTokenExpiration,
   handleQuotaCheck,
 } from './retry-handler';
+import { MANAGED_QUOTA_PROVIDERS, type ManagedQuotaProvider } from '../quota-manager';
 import { checkOrJoinProxy, registerProxySession, setupCleanupHandlers } from './session-bridge';
 import { parseThinkingOverride } from './thinking-arg-parser';
 import {
@@ -128,6 +129,25 @@ import {
 } from './thinking-override-resolver';
 import { shouldStartHttpsTunnel } from './https-tunnel-policy';
 
+function resolveRuntimeQuotaMonitorProviders(
+  provider: CLIProxyProvider,
+  compositeProviders: CLIProxyProvider[]
+): ManagedQuotaProvider[] {
+  const candidates = compositeProviders.length > 0 ? compositeProviders : [provider];
+  const resolved: ManagedQuotaProvider[] = [];
+
+  for (const candidate of candidates) {
+    if (
+      MANAGED_QUOTA_PROVIDERS.includes(candidate as ManagedQuotaProvider) &&
+      !resolved.includes(candidate as ManagedQuotaProvider)
+    ) {
+      resolved.push(candidate as ManagedQuotaProvider);
+    }
+  }
+
+  return resolved;
+}
+
 /** Default executor configuration */
 const DEFAULT_CONFIG: ExecutorConfig = {
   port: CLIPROXY_DEFAULT_PORT,
@@ -888,8 +908,7 @@ export async function execClaudeWithCLIProxy(
   if (!skipLocalAuth) {
     // Multi-tier quota check for composite variants (check if any tier uses a managed provider)
     if (compositeProviders.length > 0) {
-      const managedQuotaProviders = ['agy', 'claude'] as const;
-      for (const managedProvider of managedQuotaProviders) {
+      for (const managedProvider of MANAGED_QUOTA_PROVIDERS) {
         if (compositeProviders.includes(managedProvider)) {
           await handleQuotaCheck(managedProvider);
         }
@@ -1376,9 +1395,14 @@ export async function execClaudeWithCLIProxy(
   // 12b. Start runtime quota monitor (adaptive polling during session)
   if (!skipLocalAuth) {
     const { startQuotaMonitor } = await import('../quota-manager');
-    const monitorAccount = getDefaultAccount(provider);
-    if (monitorAccount) {
-      startQuotaMonitor(provider, monitorAccount.id);
+    for (const monitorProvider of resolveRuntimeQuotaMonitorProviders(
+      provider,
+      compositeProviders
+    )) {
+      const monitorAccount = getDefaultAccount(monitorProvider);
+      if (monitorAccount) {
+        startQuotaMonitor(monitorProvider, monitorAccount.id);
+      }
     }
   }
 
@@ -1397,4 +1421,8 @@ export async function execClaudeWithCLIProxy(
 // Re-export utility functions for backwards compatibility
 export { isPortAvailable, findAvailablePort } from './lifecycle-manager';
 
+export const __testExports = {
+  resolveRuntimeQuotaMonitorProviders,
+};
+
 export default execClaudeWithCLIProxy;

src/cliproxy/executor/retry-handler.ts

@@ -80,9 +80,9 @@ export async function handleTokenExpiration(
  * Handle quota check and auto-switching for providers with quota-based rotation.
  */
 export async function handleQuotaCheck(provider: CLIProxyProvider): Promise<void> {
-  if (provider !== 'agy' && provider !== 'claude') return;
+  const { isManagedQuotaProvider, preflightCheck } = await import('../quota-manager');
+  if (!isManagedQuotaProvider(provider)) return;
 
-  const { preflightCheck } = await import('../quota-manager');
   const preflight = await preflightCheck(provider);
 
   if (!preflight.proceed) {

src/cliproxy/quota-fetcher-ghcp.ts

@@ -48,29 +48,41 @@ interface TokenData {
 }
 
 function normalizeSnapshot(raw?: RawGhcpQuotaSnapshot): GhcpQuotaSnapshot {
+  const unlimited = Boolean(raw?.unlimited);
+  const hasPercentSignal =
+    typeof raw?.percent_remaining === 'number' && Number.isFinite(raw.percent_remaining);
+  const rawRemaining = typeof raw?.remaining === 'number' ? raw.remaining : raw?.quota_remaining;
+  const hasEntitlementRemainingSignal =
+    typeof raw?.entitlement === 'number' &&
+    Number.isFinite(raw.entitlement) &&
+    raw.entitlement > 0 &&
+    typeof rawRemaining === 'number' &&
+    Number.isFinite(rawRemaining);
+  const reported =
+    raw !== undefined && (unlimited || hasPercentSignal || hasEntitlementRemainingSignal);
   const entitlement = Number(raw?.entitlement ?? 0);
-  const remainingRaw = raw?.remaining ?? raw?.quota_remaining ?? 0;
+  const remainingRaw = rawRemaining ?? 0;
   const remaining = Number(remainingRaw);
   const safeEntitlement = Number.isFinite(entitlement) ? Math.max(0, entitlement) : 0;
   const safeRemaining = Number.isFinite(remaining) ? Math.max(0, remaining) : 0;
   const used = Math.max(0, safeEntitlement - safeRemaining);
 
-  const percentRemainingRaw =
-    typeof raw?.percent_remaining === 'number' ? raw.percent_remaining : null;
-  const percentRemaining =
-    percentRemainingRaw !== null
-      ? clampPercent(percentRemainingRaw)
-      : safeEntitlement > 0
+  const percentRemaining = unlimited
+    ? 100
+    : hasPercentSignal
+      ? clampPercent(raw.percent_remaining as number)
+      : reported && safeEntitlement > 0
         ? clampPercent((safeRemaining / safeEntitlement) * 100)
         : 0;
 
   return {
+    reported,
     entitlement: safeEntitlement,
     remaining: safeRemaining,
     used,
     percentRemaining,
     percentUsed: clampPercent(100 - percentRemaining),
-    unlimited: Boolean(raw?.unlimited),
+    unlimited,
     overageCount:
       typeof raw?.overage_count === 'number' && Number.isFinite(raw.overage_count)
         ? Math.max(0, raw.overage_count)

src/cliproxy/quota-manager.ts

@@ -15,7 +15,15 @@
 import { CLIProxyProvider } from './types';
 import { QuotaResult, fetchAccountQuota } from './quota-fetcher';
 import { fetchClaudeQuota } from './quota-fetcher-claude';
-import type { ClaudeQuotaResult } from './quota-types';
+import { fetchCodexQuota } from './quota-fetcher-codex';
+import { fetchGeminiCliQuota } from './quota-fetcher-gemini-cli';
+import { fetchGhcpQuota } from './quota-fetcher-ghcp';
+import type {
+  ClaudeQuotaResult,
+  CodexQuotaResult,
+  GeminiCliQuotaResult,
+  GhcpQuotaResult,
+} from './quota-types';
 import {
   getDefaultAccount,
   getProviderAccounts,
@@ -27,12 +35,25 @@ import {
 import { loadOrCreateUnifiedConfig } from '../config/unified-config-loader';
 import type { RuntimeMonitorConfig } from '../config/unified-config-types';
 
-type ManagedQuotaProvider = 'agy' | 'claude';
-type ManagedQuotaResult = QuotaResult | ClaudeQuotaResult;
+export type ManagedQuotaProvider = 'agy' | 'claude' | 'codex' | 'gemini' | 'ghcp';
+type ManagedQuotaResult =
+  | QuotaResult
+  | ClaudeQuotaResult
+  | CodexQuotaResult
+  | GeminiCliQuotaResult
+  | GhcpQuotaResult;
 
-const MANAGED_QUOTA_PROVIDERS: readonly ManagedQuotaProvider[] = ['agy', 'claude'];
+export const MANAGED_QUOTA_PROVIDERS: readonly ManagedQuotaProvider[] = [
+  'agy',
+  'claude',
+  'codex',
+  'gemini',
+  'ghcp',
+];
 
-function isManagedQuotaProvider(provider: CLIProxyProvider): provider is ManagedQuotaProvider {
+export function isManagedQuotaProvider(
+  provider: CLIProxyProvider
+): provider is ManagedQuotaProvider {
   return MANAGED_QUOTA_PROVIDERS.includes(provider as ManagedQuotaProvider);
 }
 
@@ -117,19 +138,7 @@ async function fetchQuotaWithDedup(
       setCachedQuota(provider, accountId, result);
       return result;
     })
-    .catch((): ManagedQuotaResult => {
-      if (provider === 'claude') {
-        return {
-          success: false,
-          windows: [],
-          coreUsage: { fiveHour: null, weekly: null },
-          lastUpdated: Date.now(),
-          error: 'Failed to fetch Claude quota',
-          accountId,
-        };
-      }
-      return { success: false, models: [], lastUpdated: Date.now() };
-    })
+    .catch((): ManagedQuotaResult => buildFailedQuotaResult(provider, accountId))
     .finally(() => {
       pendingFetches.delete(key);
     });
@@ -143,10 +152,85 @@ async function fetchManagedQuota(
   accountId: string,
   verbose: boolean
 ): Promise<ManagedQuotaResult> {
-  if (provider === 'claude') {
-    return fetchClaudeQuota(accountId, verbose);
+  switch (provider) {
+    case 'agy':
+      return fetchAccountQuota(provider, accountId, verbose);
+    case 'claude':
+      return fetchClaudeQuota(accountId, verbose);
+    case 'codex':
+      return fetchCodexQuota(accountId, verbose);
+    case 'gemini':
+      return fetchGeminiCliQuota(accountId, verbose);
+    case 'ghcp':
+      return fetchGhcpQuota(accountId, verbose);
+  }
+}
+
+function buildFailedGhcpSnapshot(): GhcpQuotaResult['snapshots']['chat'] {
+  return {
+    reported: false,
+    entitlement: 0,
+    remaining: 0,
+    used: 0,
+    percentRemaining: 0,
+    percentUsed: 0,
+    unlimited: false,
+    overageCount: 0,
+    overagePermitted: false,
+    quotaId: null,
+  };
+}
+
+function buildFailedQuotaResult(
+  provider: ManagedQuotaProvider,
+  accountId: string
+): ManagedQuotaResult {
+  switch (provider) {
+    case 'agy':
+      return { success: false, models: [], lastUpdated: Date.now(), accountId };
+    case 'claude':
+      return {
+        success: false,
+        windows: [],
+        coreUsage: { fiveHour: null, weekly: null },
+        lastUpdated: Date.now(),
+        error: 'Failed to fetch Claude quota',
+        accountId,
+      };
+    case 'codex':
+      return {
+        success: false,
+        windows: [],
+        coreUsage: { fiveHour: null, weekly: null },
+        planType: null,
+        lastUpdated: Date.now(),
+        error: 'Failed to fetch Codex quota',
+        accountId,
+      };
+    case 'gemini':
+      return {
+        success: false,
+        buckets: [],
+        projectId: null,
+        lastUpdated: Date.now(),
+        error: 'Failed to fetch Gemini quota',
+        accountId,
+      };
+    case 'ghcp':
+      return {
+        success: false,
+        planType: null,
+        quotaResetDate: null,
+        snapshots: {
+          premiumInteractions: buildFailedGhcpSnapshot(),
+          chat: buildFailedGhcpSnapshot(),
+          completions: buildFailedGhcpSnapshot(),
+        },
+        lastUpdated: Date.now(),
+        error: 'Failed to fetch GitHub Copilot quota',
+        accountId,
+      };
   }
-  return fetchAccountQuota(provider, accountId, verbose);
 }
 
 // ============================================================================
@@ -264,14 +348,62 @@ function calculateClaudeQuotaPercent(quota: ClaudeQuotaResult): number | null {
   return null;
 }
 
+function calculateCodexQuotaPercent(quota: CodexQuotaResult): number | null {
+  if (!quota.success) return null;
+
+  const coreWindows = [quota.coreUsage?.fiveHour, quota.coreUsage?.weekly].filter(
+    (window): window is NonNullable<typeof window> => !!window
+  );
+  if (coreWindows.length > 0) {
+    return Math.min(...coreWindows.map((window) => window.remainingPercent));
+  }
+
+  const usageWindows = quota.windows.filter((window) => window.category !== 'code-review');
+  if (usageWindows.length > 0) {
+    return Math.min(...usageWindows.map((window) => window.remainingPercent));
+  }
+
+  return null;
+}
+
+function calculateGeminiQuotaPercent(quota: GeminiCliQuotaResult): number | null {
+  if (!quota.success || quota.buckets.length === 0) return null;
+  return Math.min(...quota.buckets.map((bucket) => bucket.remainingPercent));
+}
+
+function calculateGhcpQuotaPercent(quota: GhcpQuotaResult): number | null {
+  if (!quota.success) return null;
+  const percentages = [
+    quota.snapshots.premiumInteractions,
+    quota.snapshots.chat,
+    quota.snapshots.completions,
+  ]
+    .filter((snapshot) => snapshot.reported !== false)
+    .map((snapshot) => (snapshot.unlimited ? 100 : snapshot.percentRemaining))
+    .filter((percentage) => Number.isFinite(percentage));
+
+  return percentages.length > 0 ? Math.min(...percentages) : null;
+}
+
 /**
  * Calculate normalized quota percentage for managed providers.
  */
-function calculateQuotaPercent(quota: ManagedQuotaResult): number | null {
-  if ('models' in quota) {
-    return calculateAgyQuotaPercent(quota);
+function calculateQuotaPercent(
+  provider: ManagedQuotaProvider,
+  quota: ManagedQuotaResult
+): number | null {
+  switch (provider) {
+    case 'agy':
+      return calculateAgyQuotaPercent(quota as QuotaResult);
+    case 'claude':
+      return calculateClaudeQuotaPercent(quota as ClaudeQuotaResult);
+    case 'codex':
+      return calculateCodexQuotaPercent(quota as CodexQuotaResult);
+    case 'gemini':
+      return calculateGeminiQuotaPercent(quota as GeminiCliQuotaResult);
+    case 'ghcp':
+      return calculateGhcpQuotaPercent(quota as GhcpQuotaResult);
   }
-  return calculateClaudeQuotaPercent(quota);
 }
 
 /**
@@ -312,7 +444,7 @@ export async function findHealthyAccount(
         quota = await fetchQuotaWithDedup(provider, account.id);
       }
 
-      const avgQuota = calculateQuotaPercent(quota);
+      const avgQuota = calculateQuotaPercent(provider, quota);
 
       return {
         id: account.id,
@@ -437,7 +569,7 @@ export async function preflightCheck(provider: CLIProxyProvider): Promise<Prefli
 
   // Calculate normalized quota percentage.
   // Null means quota data is unavailable (e.g. provider does not expose limits for this account).
-  const quotaPercent = calculateQuotaPercent(quota);
+  const quotaPercent = calculateQuotaPercent(provider, quota);
   if (quotaPercent === null) {
     return {
       proceed: true,
@@ -462,11 +594,13 @@ export async function preflightCheck(provider: CLIProxyProvider): Promise<Prefli
       };
     }
 
-    pauseAccountForQuotaCooldown(
-      provider,
-      defaultAccount.id,
-      quotaConfig.auto?.cooldown_minutes ?? 5
-    );
+    if (alternative.lastQuota !== null && alternative.lastQuota >= threshold) {
+      pauseAccountForQuotaCooldown(
+        provider,
+        defaultAccount.id,
+        quotaConfig.auto?.cooldown_minutes ?? 5
+      );
+    }
     setDefaultAccount(provider, alternative.id);
     touchAccount(provider, alternative.id);
     return {
@@ -506,12 +640,14 @@ export async function getQuotaStatus(provider: CLIProxyProvider): Promise<{
 
   const results = await Promise.all(
     accounts.map(async (account) => {
+      const managedProvider = isManagedQuotaProvider(provider) ? provider : null;
       let quota = getCachedQuota(provider, account.id);
-      if (!quota && isManagedQuotaProvider(provider)) {
-        quota = await fetchQuotaWithDedup(provider, account.id);
+      if (!quota && managedProvider) {
+        quota = await fetchQuotaWithDedup(managedProvider, account.id);
       }
 
-      const avgQuota = quota ? calculateQuotaPercent(quota) : null;
+      const avgQuota =
+        quota && managedProvider ? calculateQuotaPercent(managedProvider, quota) : null;
 
       return {
         account,
@@ -530,14 +666,14 @@ export async function getQuotaStatus(provider: CLIProxyProvider): Promise<{
 // RUNTIME QUOTA MONITOR (adaptive polling during active sessions)
 // ============================================================================
 
-/** Active monitor timer (null = not running) */
-let monitorTimer: ReturnType<typeof setTimeout> | null = null;
+interface RuntimeMonitorState {
+  timer: ReturnType<typeof setTimeout> | null;
+  hasWarnedThisSession: boolean;
+  stopped: boolean;
+}
 
-/** Tracks if warning was shown this session (avoid spam) */
-let hasWarnedThisSession = false;
-
-/** Guards against in-flight poll callbacks running after stop */
-let monitorStopped = false;
+/** Active monitor timers keyed by provider/account. */
+const activeRuntimeMonitors = new Map<string, RuntimeMonitorState>();
 
 /**
  * Schedule next quota poll with adaptive interval.
@@ -547,16 +683,18 @@ function scheduleNextPoll(
   provider: ManagedQuotaProvider,
   accountId: string,
   monitorConfig: RuntimeMonitorConfig,
-  intervalMs: number
+  intervalMs: number,
+  state: RuntimeMonitorState
 ): void {
-  monitorTimer = setTimeout(async () => {
+  const monitorKey = getCacheKey(provider, accountId);
+  state.timer = setTimeout(async () => {
     // Guard: skip if monitor was stopped while this callback was queued
-    if (monitorStopped) return;
+    if (state.stopped || activeRuntimeMonitors.get(monitorKey) !== state) return;
 
     try {
       const quota = await fetchQuotaWithDedup(provider, accountId);
-      if (monitorStopped) return; // Re-check after async fetch
-      const avgQuota = calculateQuotaPercent(quota);
+      if (state.stopped || activeRuntimeMonitors.get(monitorKey) !== state) return;
+      const avgQuota = calculateQuotaPercent(provider, quota);
 
       if (avgQuota === null) {
         // Quota data unavailable: keep polling, but do not treat unknown as healthy/exhausted.
@@ -564,7 +702,8 @@ function scheduleNextPoll(
           provider,
           accountId,
           monitorConfig,
-          monitorConfig.normal_interval_seconds * 1000
+          monitorConfig.normal_interval_seconds * 1000,
+          state
         );
         return;
       }
@@ -576,22 +715,25 @@ function scheduleNextPoll(
         // default only takes effect on next session start via preflightCheck().
         const { handleQuotaExhaustion } = await import('./account-safety');
         await handleQuotaExhaustion(provider, accountId, monitorConfig.cooldown_minutes);
-        monitorTimer = null;
+        state.stopped = true;
+        state.timer = null;
+        activeRuntimeMonitors.delete(monitorKey);
         return; // Stop polling
       }
 
       if (avgQuota <= monitorConfig.warn_threshold) {
         // WARNING: switch to critical interval, warn once
-        if (!hasWarnedThisSession) {
+        if (!state.hasWarnedThisSession) {
           const { writeQuotaWarning } = await import('./account-safety');
           writeQuotaWarning(accountId, avgQuota);
-          hasWarnedThisSession = true;
+          state.hasWarnedThisSession = true;
         }
         scheduleNextPoll(
           provider,
           accountId,
           monitorConfig,
-          monitorConfig.critical_interval_seconds * 1000
+          monitorConfig.critical_interval_seconds * 1000,
+          state
         );
         return;
       }
@@ -601,17 +743,20 @@ function scheduleNextPoll(
         provider,
         accountId,
         monitorConfig,
-        monitorConfig.normal_interval_seconds * 1000
+        monitorConfig.normal_interval_seconds * 1000,
+        state
       );
     } catch {
       // API failure: silently reschedule at same interval
-      scheduleNextPoll(provider, accountId, monitorConfig, intervalMs);
+      if (!state.stopped && activeRuntimeMonitors.get(monitorKey) === state) {
+        scheduleNextPoll(provider, accountId, monitorConfig, intervalMs, state);
+      }
     }
   }, intervalMs);
 
   // Prevent monitor from keeping Node.js process alive
-  if (monitorTimer && typeof monitorTimer === 'object' && 'unref' in monitorTimer) {
-    monitorTimer.unref();
+  if (state.timer && typeof state.timer === 'object' && 'unref' in state.timer) {
+    state.timer.unref();
   }
 }
 
@@ -628,9 +773,6 @@ export function startQuotaMonitor(provider: CLIProxyProvider, accountId: string)
   // Only managed providers support runtime quota monitoring.
   if (!isManagedQuotaProvider(provider)) return;
 
-  // Prevent duplicate monitors
-  if (monitorTimer) return;
-
   const config = loadOrCreateUnifiedConfig();
   const quotaConfig = config.quota_management;
 
@@ -647,15 +789,23 @@ export function startQuotaMonitor(provider: CLIProxyProvider, accountId: string)
     return; // Invalid config — skip monitoring silently (logged at config level)
   }
 
-  hasWarnedThisSession = false;
-  monitorStopped = false;
+  const monitorKey = getCacheKey(provider, accountId);
+  if (activeRuntimeMonitors.has(monitorKey)) return;
+
+  const state: RuntimeMonitorState = {
+    timer: null,
+    hasWarnedThisSession: false,
+    stopped: false,
+  };
+  activeRuntimeMonitors.set(monitorKey, state);
 
   // Start first poll at normal interval
   scheduleNextPoll(
     provider,
     accountId,
     quotaConfig.runtime_monitor,
-    quotaConfig.runtime_monitor.normal_interval_seconds * 1000
+    quotaConfig.runtime_monitor.normal_interval_seconds * 1000,
+    state
   );
 }
 
@@ -663,10 +813,12 @@ export function startQuotaMonitor(provider: CLIProxyProvider, accountId: string)
  * Stop the runtime quota monitor. Safe to call multiple times.
  */
 export function stopQuotaMonitor(): void {
-  monitorStopped = true;
-  if (monitorTimer) {
-    clearTimeout(monitorTimer);
-    monitorTimer = null;
+  for (const state of activeRuntimeMonitors.values()) {
+    state.stopped = true;
+    if (state.timer) {
+      clearTimeout(state.timer);
+      state.timer = null;
+    }
   }
-  hasWarnedThisSession = false;
+  activeRuntimeMonitors.clear();
 }

src/cliproxy/quota-types.ts

@@ -230,6 +230,8 @@ export interface GeminiCliQuotaResult extends QuotaErrorMetadata {
  * GitHub Copilot quota snapshot.
  */
 export interface GhcpQuotaSnapshot {
+  /** False when upstream omitted this quota category; callers should exclude it from health math */
+  reported?: boolean;
   /** Total quota allocation for this category */
   entitlement: number;
   /** Remaining quota count */

src/commands/cliproxy/quota-subcommand.ts

@@ -752,13 +752,17 @@ function displayGhcpQuotaSection(results: { account: string; quota: GhcpQuotaRes
       continue;
     }
 
-    const rows = [
-      quota.snapshots.premiumInteractions.percentRemaining,
-      quota.snapshots.chat.percentRemaining,
-      quota.snapshots.completions.percentRemaining,
-    ];
-    const minQuota = rows.length > 0 ? Math.min(...rows) : 0;
-    const statusIcon = minQuota > 50 ? ok('') : minQuota > 10 ? warn('') : fail('');
+    const reportedSnapshots = [
+      quota.snapshots.premiumInteractions,
+      quota.snapshots.chat,
+      quota.snapshots.completions,
+    ].filter((snapshot) => snapshot.reported !== false);
+    const rows = reportedSnapshots.map((snapshot) =>
+      snapshot.unlimited ? 100 : snapshot.percentRemaining
+    );
+    const minQuota = rows.length > 0 ? Math.min(...rows) : null;
+    const statusIcon =
+      minQuota === null ? info('') : minQuota > 50 ? ok('') : minQuota > 10 ? warn('') : fail('');
     const planBadge = quota.planType ? color(` [${quota.planType}]`, 'info') : '';
 
     console.log(`  ${statusIcon}${accountLabel}${defaultMark}${planBadge}`);
@@ -766,12 +770,14 @@ function displayGhcpQuotaSection(results: { account: string; quota: GhcpQuotaRes
       console.log(`    ${dim(`Resets ${formatResetTimeISO(quota.quotaResetDate)}`)}`);
     }
 
-    const items: Array<[string, GhcpQuotaResult['snapshots'][keyof GhcpQuotaResult['snapshots']]]> =
-      [
-        ['Premium interactions', quota.snapshots.premiumInteractions],
-        ['Chat', quota.snapshots.chat],
-        ['Completions', quota.snapshots.completions],
-      ];
+    const allItems: Array<
+      [string, GhcpQuotaResult['snapshots'][keyof GhcpQuotaResult['snapshots']]]
+    > = [
+      ['Premium interactions', quota.snapshots.premiumInteractions],
+      ['Chat', quota.snapshots.chat],
+      ['Completions', quota.snapshots.completions],
+    ];
+    const items = allItems.filter(([, snapshot]) => snapshot.reported !== false);
 
     for (const [label, snapshot] of items) {
       const bar = formatQuotaBar(snapshot.percentRemaining);

tests/unit/cliproxy/account-safety-quota-exhaustion.test.ts

@@ -86,6 +86,36 @@ function writeClaudeAuth(accountId: string, accessToken: string): void {
   );
 }
 
+function writeCodexAuth(tokenFile: string, accountId: string, accessToken: string): void {
+  writeAuthToken(tokenFile, {
+    access_token: accessToken,
+    account_id: `chatgpt-${accountId}`,
+    expired: '2099-01-01T00:00:00.000Z',
+    type: 'codex',
+    email: accountId,
+  });
+}
+
+function writeGhcpAuth(tokenFile: string, accessToken: string): void {
+  writeAuthToken(tokenFile, {
+    access_token: accessToken,
+    type: 'ghcp',
+  });
+}
+
+function writeGeminiAuth(tokenFile: string, accountId: string, accessToken: string): void {
+  writeAuthToken(tokenFile, {
+    type: 'gemini',
+    email: accountId,
+    project_id: 'cloudaicompanion-test-123',
+    token: {
+      access_token: accessToken,
+      refresh_token: `${accessToken}-refresh`,
+      expiry: Date.now() + 60 * 60 * 1000,
+    },
+  });
+}
+
 function writeAuthToken(tokenFile: string, payload: Record<string, unknown>): void {
   const authDir = path.join(tmpDir, '.ccs', 'cliproxy', 'auth');
   fs.mkdirSync(authDir, { recursive: true });
@@ -264,7 +294,7 @@ describe('Quota Exhaustion Handlers', () => {
       expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(false);
     });
 
-    it('should switch Claude accounts when fallback quota endpoint returns 404', async () => {
+    it('should switch Claude accounts without durable pause when fallback quota is unknown', async () => {
       writeRegistry({
         claude: {
           default: 'exhausted@example.com',
@@ -312,11 +342,27 @@ describe('Quota Exhaustion Handlers', () => {
       }) as typeof fetch;
 
       const result = await handleQuotaExhaustion('claude', 'exhausted@example.com', 10);
-      const { getDefaultAccount } = await import('../../../src/cliproxy/account-manager');
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
 
       expect(result.switchedTo).toBe('fallback@example.com');
       expect(getDefaultAccount('claude')?.id).toBe('fallback@example.com');
-      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(true);
+      expect(getAccount('claude', 'exhausted@example.com')?.paused).not.toBe(true);
+      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(
+        false
+      );
+      expect(
+        fs.existsSync(
+          path.join(
+            tmpDir,
+            '.ccs',
+            'cliproxy',
+            'auth',
+            `claude-${sanitizeEmail('exhausted@example.com')}.json`
+          )
+        )
+      ).toBe(true);
       expect(
         fs.existsSync(
           path.join(
@@ -327,6 +373,512 @@ describe('Quota Exhaustion Handlers', () => {
             `claude-${sanitizeEmail('exhausted@example.com')}.json`
           )
         )
+      ).toBe(false);
+    });
+
+    it('should self-pause exhausted Codex accounts when a healthy fallback exists', async () => {
+      writeRegistry({
+        codex: {
+          default: 'exhausted@example.com',
+          accounts: {
+            'exhausted@example.com': {
+              email: 'exhausted@example.com',
+              tokenFile: 'codex-exhausted.json',
+            },
+            'fallback@example.com': {
+              email: 'fallback@example.com',
+              tokenFile: 'codex-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeCodexAuth('codex-exhausted.json', 'exhausted@example.com', 'exhausted-token');
+      writeCodexAuth('codex-fallback.json', 'fallback@example.com', 'fallback-token');
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const accountHeader = new Headers(options?.headers).get('ChatGPT-Account-Id') ?? '';
+        const usedPercent = accountHeader === 'chatgpt-fallback@example.com' ? 10 : 100;
+        return Promise.resolve(
+          Response.json({
+            plan_type: 'pro',
+            rate_limit: {
+              primary_window: { used_percent: usedPercent, reset_after_seconds: 3600 },
+              secondary_window: { used_percent: usedPercent, reset_after_seconds: 604800 },
+            },
+          })
+        );
+      }) as typeof fetch;
+
+      const result = await handleQuotaExhaustion('codex', 'exhausted@example.com', 10);
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedTo).toBe('fallback@example.com');
+      expect(getDefaultAccount('codex')?.id).toBe('fallback@example.com');
+      expect(getAccount('codex', 'exhausted@example.com')?.paused).toBe(true);
+      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(true);
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'codex-exhausted.json'))
+      ).toBe(true);
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth', 'codex-exhausted.json'))
+      ).toBe(false);
+    });
+
+    it('should not durably self-pause the only exhausted Codex account during preflight', async () => {
+      writeRegistry({
+        codex: {
+          default: 'only-codex@example.com',
+          accounts: {
+            'only-codex@example.com': {
+              email: 'only-codex@example.com',
+              tokenFile: 'codex-only.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeCodexAuth('codex-only.json', 'only-codex@example.com', 'only-token');
+
+      global.fetch = mock(() =>
+        Promise.resolve(
+          Response.json({
+            plan_type: 'pro',
+            rate_limit: {
+              primary_window: { used_percent: 100, reset_after_seconds: 3600 },
+              secondary_window: { used_percent: 100, reset_after_seconds: 604800 },
+            },
+          })
+        )
+      ) as typeof fetch;
+
+      const { preflightCheck } = await import('../../../src/cliproxy/quota-manager');
+      const result = await preflightCheck('codex');
+      const { getAccount } = await import('../../../src/cliproxy/account-manager');
+
+      expect(result.accountId).toBe('only-codex@example.com');
+      expect(result.switchedFrom).toBeUndefined();
+      expect(result.reason).toContain('no alternatives available');
+      expect(getAccount('codex', 'only-codex@example.com')?.paused).not.toBe(true);
+      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth', 'codex-only.json'))).toBe(
+        true
+      );
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'codex-only.json'))
+      ).toBe(false);
+      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(false);
+    });
+
+    it('should self-pause an exhausted Codex default during preflight when fallback exists', async () => {
+      writeRegistry({
+        codex: {
+          default: 'preflight-exhausted@example.com',
+          accounts: {
+            'preflight-exhausted@example.com': {
+              email: 'preflight-exhausted@example.com',
+              tokenFile: 'codex-preflight-exhausted.json',
+            },
+            'preflight-fallback@example.com': {
+              email: 'preflight-fallback@example.com',
+              tokenFile: 'codex-preflight-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeCodexAuth(
+        'codex-preflight-exhausted.json',
+        'preflight-exhausted@example.com',
+        'preflight-exhausted-token'
+      );
+      writeCodexAuth(
+        'codex-preflight-fallback.json',
+        'preflight-fallback@example.com',
+        'preflight-fallback-token'
+      );
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const accountHeader = new Headers(options?.headers).get('ChatGPT-Account-Id') ?? '';
+        const usedPercent = accountHeader === 'chatgpt-preflight-fallback@example.com' ? 10 : 100;
+        return Promise.resolve(
+          Response.json({
+            plan_type: 'pro',
+            rate_limit: {
+              primary_window: { used_percent: usedPercent, reset_after_seconds: 3600 },
+              secondary_window: { used_percent: usedPercent, reset_after_seconds: 604800 },
+            },
+          })
+        );
+      }) as typeof fetch;
+
+      const { preflightCheck } = await import('../../../src/cliproxy/quota-manager');
+      const result = await preflightCheck('codex');
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedFrom).toBe('preflight-exhausted@example.com');
+      expect(result.accountId).toBe('preflight-fallback@example.com');
+      expect(getDefaultAccount('codex')?.id).toBe('preflight-fallback@example.com');
+      expect(getAccount('codex', 'preflight-exhausted@example.com')?.paused).toBe(true);
+      expect(
+        fs.existsSync(
+          path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'codex-preflight-exhausted.json')
+        )
+      ).toBe(true);
+    });
+
+    it('should not durably self-pause during preflight when fallback quota is unknown', async () => {
+      writeRegistry({
+        codex: {
+          default: 'preflight-unknown-exhausted@example.com',
+          accounts: {
+            'preflight-unknown-exhausted@example.com': {
+              email: 'preflight-unknown-exhausted@example.com',
+              tokenFile: 'codex-preflight-unknown-exhausted.json',
+            },
+            'preflight-unknown-fallback@example.com': {
+              email: 'preflight-unknown-fallback@example.com',
+              tokenFile: 'codex-preflight-unknown-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeCodexAuth(
+        'codex-preflight-unknown-exhausted.json',
+        'preflight-unknown-exhausted@example.com',
+        'preflight-unknown-exhausted-token'
+      );
+      writeCodexAuth(
+        'codex-preflight-unknown-fallback.json',
+        'preflight-unknown-fallback@example.com',
+        'preflight-unknown-fallback-token'
+      );
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const accountHeader = new Headers(options?.headers).get('ChatGPT-Account-Id') ?? '';
+        if (accountHeader === 'chatgpt-preflight-unknown-fallback@example.com') {
+          return Promise.resolve(new Response('', { status: 500 }));
+        }
+
+        return Promise.resolve(
+          Response.json({
+            plan_type: 'pro',
+            rate_limit: {
+              primary_window: { used_percent: 100, reset_after_seconds: 3600 },
+              secondary_window: { used_percent: 100, reset_after_seconds: 604800 },
+            },
+          })
+        );
+      }) as typeof fetch;
+
+      const { preflightCheck } = await import('../../../src/cliproxy/quota-manager');
+      const result = await preflightCheck('codex');
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedFrom).toBe('preflight-unknown-exhausted@example.com');
+      expect(result.accountId).toBe('preflight-unknown-fallback@example.com');
+      expect(getDefaultAccount('codex')?.id).toBe('preflight-unknown-fallback@example.com');
+      expect(getAccount('codex', 'preflight-unknown-exhausted@example.com')?.paused).not.toBe(
+        true
+      );
+      expect(
+        fs.existsSync(
+          path.join(tmpDir, '.ccs', 'cliproxy', 'auth', 'codex-preflight-unknown-exhausted.json')
+        )
+      ).toBe(true);
+      expect(
+        fs.existsSync(
+          path.join(
+            tmpDir,
+            '.ccs',
+            'cliproxy',
+            'auth-paused',
+            'codex-preflight-unknown-exhausted.json'
+          )
+        )
+      ).toBe(false);
+      expect(fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'quota-paused.json'))).toBe(
+        false
+      );
+    });
+
+    it('should self-pause exhausted Gemini accounts when a healthy fallback exists', async () => {
+      writeRegistry({
+        gemini: {
+          default: 'gemini-exhausted@example.com',
+          accounts: {
+            'gemini-exhausted@example.com': {
+              email: 'gemini-exhausted@example.com',
+              tokenFile: 'gemini-exhausted.json',
+            },
+            'gemini-fallback@example.com': {
+              email: 'gemini-fallback@example.com',
+              tokenFile: 'gemini-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeGeminiAuth('gemini-exhausted.json', 'gemini-exhausted@example.com', 'exhausted-token');
+      writeGeminiAuth('gemini-fallback.json', 'gemini-fallback@example.com', 'fallback-token');
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const authHeader = new Headers(options?.headers).get('Authorization') ?? '';
+        const remainingFraction = authHeader === 'Bearer fallback-token' ? 0.9 : 0;
+        return Promise.resolve(
+          Response.json({
+            buckets: [
+              {
+                model_id: 'gemini-3-flash-preview',
+                remaining_fraction: remainingFraction,
+                remaining_amount: Math.round(remainingFraction * 100),
+                reset_time: '2026-05-01T00:00:00Z',
+              },
+            ],
+          })
+        );
+      }) as typeof fetch;
+
+      const result = await handleQuotaExhaustion('gemini', 'gemini-exhausted@example.com', 10);
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedTo).toBe('gemini-fallback@example.com');
+      expect(getDefaultAccount('gemini')?.id).toBe('gemini-fallback@example.com');
+      expect(getAccount('gemini', 'gemini-exhausted@example.com')?.paused).toBe(true);
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'gemini-exhausted.json'))
+      ).toBe(true);
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth', 'gemini-exhausted.json'))
+      ).toBe(false);
+    });
+
+    it('should self-pause exhausted GitHub Copilot accounts when a healthy fallback exists', async () => {
+      writeRegistry({
+        ghcp: {
+          default: 'ghcp-exhausted',
+          accounts: {
+            'ghcp-exhausted': {
+              tokenFile: 'ghcp-exhausted.json',
+            },
+            'ghcp-fallback': {
+              tokenFile: 'ghcp-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeGhcpAuth('ghcp-exhausted.json', 'exhausted-token');
+      writeGhcpAuth('ghcp-fallback.json', 'fallback-token');
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const authHeader = new Headers(options?.headers).get('Authorization') ?? '';
+        const remaining = authHeader === 'token fallback-token' ? 90 : 0;
+        return Promise.resolve(
+          Response.json({
+            copilot_plan: 'individual',
+            quota_reset_date: '2026-05-01T00:00:00Z',
+            quota_snapshots: {
+              premium_interactions: { entitlement: 100, remaining },
+              chat: { entitlement: 100, remaining },
+              completions: { entitlement: 100, remaining },
+            },
+          })
+        );
+      }) as typeof fetch;
+
+      const result = await handleQuotaExhaustion('ghcp', 'ghcp-exhausted', 10);
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedTo).toBe('ghcp-fallback');
+      expect(getDefaultAccount('ghcp')?.id).toBe('ghcp-fallback');
+      expect(getAccount('ghcp', 'ghcp-exhausted')?.paused).toBe(true);
+      expect(
+        fs.existsSync(path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'ghcp-exhausted.json'))
+      ).toBe(true);
+    });
+
+    it('should ignore omitted GitHub Copilot snapshots when selecting a fallback', async () => {
+      writeRegistry({
+        ghcp: {
+          default: 'ghcp-partial-exhausted',
+          accounts: {
+            'ghcp-partial-exhausted': {
+              tokenFile: 'ghcp-partial-exhausted.json',
+            },
+            'ghcp-partial-fallback': {
+              tokenFile: 'ghcp-partial-fallback.json',
+            },
+          },
+        },
+      });
+
+      writeConfig({
+        mode: 'auto',
+        auto: {
+          tier_priority: ['ultra', 'pro', 'free'],
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+          preflight_check: true,
+        },
+        runtime_monitor: {
+          enabled: true,
+          normal_interval_seconds: 300,
+          critical_interval_seconds: 60,
+          warn_threshold: 20,
+          exhaustion_threshold: 5,
+          cooldown_minutes: 10,
+        },
+      });
+
+      writeGhcpAuth('ghcp-partial-exhausted.json', 'partial-exhausted-token');
+      writeGhcpAuth('ghcp-partial-fallback.json', 'partial-fallback-token');
+
+      global.fetch = mock((_url: string, options?: RequestInit) => {
+        const authHeader = new Headers(options?.headers).get('Authorization') ?? '';
+        const remaining = authHeader === 'token partial-fallback-token' ? 90 : 0;
+        return Promise.resolve(
+          Response.json({
+            copilot_plan: 'individual',
+            quota_reset_date: '2026-05-01T00:00:00Z',
+            quota_snapshots: {
+              premium_interactions: { entitlement: 100, remaining },
+              chat: { entitlement: 100, remaining },
+            },
+          })
+        );
+      }) as typeof fetch;
+
+      const result = await handleQuotaExhaustion('ghcp', 'ghcp-partial-exhausted', 10);
+      const { getAccount, getDefaultAccount } = await import(
+        '../../../src/cliproxy/account-manager'
+      );
+
+      expect(result.switchedTo).toBe('ghcp-partial-fallback');
+      expect(getDefaultAccount('ghcp')?.id).toBe('ghcp-partial-fallback');
+      expect(getAccount('ghcp', 'ghcp-partial-exhausted')?.paused).toBe(true);
+      expect(
+        fs.existsSync(
+          path.join(tmpDir, '.ccs', 'cliproxy', 'auth-paused', 'ghcp-partial-exhausted.json')
+        )
       ).toBe(true);
     });
 

tests/unit/cliproxy/quota-fetcher-ghcp.test.ts

@@ -76,6 +76,7 @@ describe('GHCP Quota Fetcher', () => {
       const snapshot = normalizeGhcpSnapshot();
 
       expect(snapshot).toEqual({
+        reported: false,
         entitlement: 0,
         remaining: 0,
         used: 0,
@@ -88,6 +89,50 @@ describe('GHCP Quota Fetcher', () => {
       });
     });
 
+    it('marks upstream-provided snapshots as reported', () => {
+      const snapshot = normalizeGhcpSnapshot({
+        entitlement: 100,
+        remaining: 80,
+      });
+
+      expect(snapshot.reported).toBe(true);
+    });
+
+    it('treats unknown upstream snapshot objects as unreported', () => {
+      const empty = normalizeGhcpSnapshot({});
+      const quotaIdOnly = normalizeGhcpSnapshot({ quota_id: 'chat' });
+      const remainingOnly = normalizeGhcpSnapshot({ remaining: 80 });
+
+      for (const snapshot of [empty, quotaIdOnly, remainingOnly]) {
+        expect(snapshot.reported).toBe(false);
+        expect(snapshot.percentRemaining).toBe(0);
+        expect(snapshot.percentUsed).toBe(100);
+      }
+    });
+
+    it('marks percent-only snapshots as reported', () => {
+      const snapshot = normalizeGhcpSnapshot({
+        percent_remaining: 70,
+      });
+
+      expect(snapshot.reported).toBe(true);
+      expect(snapshot.percentRemaining).toBe(70);
+      expect(snapshot.percentUsed).toBe(30);
+    });
+
+    it('treats unlimited snapshots as 100% remaining', () => {
+      const snapshot = normalizeGhcpSnapshot({
+        unlimited: true,
+        entitlement: 0,
+        remaining: 0,
+      });
+
+      expect(snapshot.reported).toBe(true);
+      expect(snapshot.percentRemaining).toBe(100);
+      expect(snapshot.percentUsed).toBe(0);
+      expect(snapshot.unlimited).toBe(true);
+    });
+
     it('clamps percent_remaining to 0-100 range', () => {
       const above = normalizeGhcpSnapshot({
         entitlement: 100,

tests/unit/cliproxy/quota-monitor-runtime.test.ts

@@ -3,7 +3,7 @@
  *
  * Tests the quota monitor lifecycle:
  * - startQuotaMonitor / stopQuotaMonitor behavior
- * - No-op conditions for non-agy, manual mode, disabled config
+ * - No-op conditions for unsupported providers, manual mode, disabled config
  * - Idempotent stopQuotaMonitor
  */
 
@@ -16,6 +16,7 @@ import {
   stopQuotaMonitor,
   clearQuotaCache,
 } from '../../../src/cliproxy/quota-manager';
+import { __testExports as executorTestExports } from '../../../src/cliproxy/executor';
 
 // Setup test isolation
 let tmpDir: string;
@@ -40,11 +41,43 @@ afterEach(() => {
 });
 
 describe('Runtime Quota Monitor', () => {
+  describe('resolveRuntimeQuotaMonitorProviders', () => {
+    it('should monitor the default provider for non-composite sessions', () => {
+      expect(executorTestExports.resolveRuntimeQuotaMonitorProviders('codex', [])).toEqual([
+        'codex',
+      ]);
+    });
+
+    it('should monitor every unique managed provider in composite sessions', () => {
+      expect(
+        executorTestExports.resolveRuntimeQuotaMonitorProviders('claude', [
+          'codex',
+          'qwen',
+          'gemini',
+          'codex',
+        ])
+      ).toEqual(['codex', 'gemini']);
+    });
+
+    it('should skip unsupported non-composite providers', () => {
+      expect(executorTestExports.resolveRuntimeQuotaMonitorProviders('qwen', [])).toEqual([]);
+    });
+  });
+
   describe('startQuotaMonitor', () => {
-    it('should accept non-agy provider without throwing', () => {
-      // Non-agy providers should be silently ignored
+    it('should accept all quota-supported providers without throwing', () => {
+      for (const provider of ['agy', 'claude', 'codex', 'gemini', 'ghcp'] as const) {
+        expect(() => {
+          startQuotaMonitor(provider, 'test@gmail.com');
+          stopQuotaMonitor();
+        }).not.toThrow();
+      }
+    });
+
+    it('should ignore unsupported providers without throwing', () => {
       expect(() => {
-        startQuotaMonitor('gemini', 'test@gmail.com');
+        startQuotaMonitor('qwen', 'test@gmail.com');
+        stopQuotaMonitor();
       }).not.toThrow();
     });
 
@@ -75,6 +108,54 @@ describe('Runtime Quota Monitor', () => {
       }).not.toThrow();
     });
 
+    it('should start independent monitors for distinct managed provider accounts', () => {
+      const configDir = path.join(tmpDir, '.ccs', 'config');
+      fs.mkdirSync(configDir, { recursive: true });
+      fs.writeFileSync(
+        path.join(configDir, 'unified-config.json'),
+        JSON.stringify({
+          version: 2,
+          quota_management: {
+            mode: 'auto',
+            runtime_monitor: {
+              enabled: true,
+              normal_interval_seconds: 300,
+              critical_interval_seconds: 60,
+              warn_threshold: 20,
+              exhaustion_threshold: 5,
+              cooldown_minutes: 5,
+            },
+          },
+        })
+      );
+
+      const originalSetTimeout = globalThis.setTimeout;
+      const fakeTimer = {
+        unref: () => fakeTimer,
+        ref: () => fakeTimer,
+        hasRef: () => false,
+        refresh: () => fakeTimer,
+      } as unknown as ReturnType<typeof setTimeout>;
+      const observedDelays: Array<number | undefined> = [];
+
+      globalThis.setTimeout = ((handler: TimerHandler, timeout?: number) => {
+        observedDelays.push(timeout);
+        void handler;
+        return fakeTimer;
+      }) as typeof globalThis.setTimeout;
+
+      try {
+        startQuotaMonitor('codex', 'codex@example.com');
+        startQuotaMonitor('gemini', 'gemini@example.com');
+        startQuotaMonitor('codex', 'codex@example.com');
+      } finally {
+        stopQuotaMonitor();
+        globalThis.setTimeout = originalSetTimeout;
+      }
+
+      expect(observedDelays).toEqual([300_000, 300_000]);
+    });
+
     it('should be no-op when config missing or no quota_management', () => {
       // No config file — should not throw
       expect(() => {

ui/src/components/shared/quota-tooltip-content.tsx

@@ -465,7 +465,7 @@ export function QuotaTooltipContent({ quota, resetTime }: QuotaTooltipContentPro
       },
       { label: t('quotaTooltip.chat'), snapshot: quota.snapshots.chat },
       { label: t('quotaTooltip.completions'), snapshot: quota.snapshots.completions },
-    ];
+    ].filter(({ snapshot }) => snapshot.reported !== false);
     const effectiveResetTime = quota.quotaResetDate ?? resetTime;
     const planLabel = formatPlanLabel(quota.planType);
 

ui/src/lib/api-client.ts

@@ -862,6 +862,8 @@ export interface GeminiCliQuotaResult {
 
 /** GitHub Copilot quota snapshot */
 export interface GhcpQuotaSnapshot {
+  /** False when upstream omitted this quota category */
+  reported?: boolean;
   /** Total quota allocation for this category */
   entitlement: number;
   /** Remaining quota count */

ui/src/lib/utils.ts

@@ -694,11 +694,10 @@ export function getGeminiResetTime(buckets: GeminiCliBucket[]): string | null {
 export function getMinGhcpQuota(snapshots: GhcpQuotaResult['snapshots']): number | null {
   if (!snapshots) return null;
 
-  const percentages = [
-    snapshots.premiumInteractions.percentRemaining,
-    snapshots.chat.percentRemaining,
-    snapshots.completions.percentRemaining,
-  ].filter((p) => typeof p === 'number' && isFinite(p));
+  const percentages = [snapshots.premiumInteractions, snapshots.chat, snapshots.completions]
+    .filter((snapshot) => snapshot.reported !== false)
+    .map((snapshot) => (snapshot.unlimited ? 100 : snapshot.percentRemaining))
+    .filter((p) => typeof p === 'number' && isFinite(p));
 
   if (percentages.length === 0) return null;
   return Math.min(...percentages);