diff --git a/changelog.d/fixes/9041-rate-limit-idle-capacity-wedge.md b/changelog.d/fixes/9041-rate-limit-idle-capacity-wedge.md
new file mode 100644
index 0000000000..cfb180cabd
--- /dev/null
+++ b/changelog.d/fixes/9041-rate-limit-idle-capacity-wedge.md
@@ -0,0 +1 @@
+- **fix(resilience):** Detect and reset idle-capacity rate-limit queue wedges on an eligible watchdog scan so routing can fall back promptly ([#9041](https://github.com/diegosouzapw/OmniRoute/pull/9041))
diff --git a/docs/architecture/RESILIENCE_GUIDE.md b/docs/architecture/RESILIENCE_GUIDE.md
index 8e68322e96..e2f9d3dfd2 100644
--- a/docs/architecture/RESILIENCE_GUIDE.md
+++ b/docs/architecture/RESILIENCE_GUIDE.md
@@ -224,12 +224,16 @@ rate limit. Bounded by `comboCooldownWait` (`enabled`, `maxWaitMs`, `maxAttempts
 **Scope**: the local per-provider+connection rate-limit queue (`open-sse/services/rateLimitManager.ts`,
 backed by Bottleneck), one layer below the three mechanisms above.
 
-**`maxWaitMs` default lowered 120s → 15s.** `resilienceSettings.requestQueue.maxWaitMs`
-bounds how long a request may wait in the local queue before it is dropped
-(`code: "RATE_LIMIT_QUEUE_TIMEOUT"`, #4165). The factory default fell from 120000ms to
-15000ms so a saturated queue fails fast instead of holding a caller for two
-minutes; override via `RATE_LIMIT_MAX_WAIT_MS` (env) or the dashboard
-(**Settings → Resilience**, 1–30000ms UI ceiling).
+**`maxWaitMs` is a legacy persisted name for execution expiration.**
+`resilienceSettings.requestQueue.maxWaitMs` is passed to Bottleneck as a job
+`expiration`, whose timer starts only after dispatch. It therefore bounds
+limiter-managed execution, not time spent in the local queue. Expiration is
+surfaced as trusted local `code: "RATE_LIMIT_EXECUTION_TIMEOUT"` (HTTP 504);
+the former queue-timeout code name is accepted only for trusted internal
+backward compatibility. The default is 15000ms; override via
+`RATE_LIMIT_MAX_WAIT_MS` (env) or the dashboard (**Settings → Resilience**,
+1–30000ms UI ceiling). Queue residence has no time deadline; use
+`maxQueueDepth` below to bound queued callers.
 
 **`maxQueueDepth` — opt-in admission cap (new).** `resilienceSettings.requestQueue.maxQueueDepth`
 bounds how many requests may sit queued (not yet dispatched) for one
@@ -252,7 +256,7 @@ it is unit-testable without a real Bottleneck limiter.
 > around the `resolveCompressionSettings`/`selectCompressionStrategy` block),
 > not HTTP response compression on synthesized 429 bodies — there is no
 > matching code path for a literal bypass flag. That prompt-compression step
-> also currently runs *before* `withRateLimit()` in the request pipeline, so
+> also currently runs _before_ `withRateLimit()` in the request pipeline, so
 > reordering to skip it on a queue-full rejection is a separate, larger
 > change than this issue's scope; it was intentionally **not** implemented
 > here and is left as a follow-up if the CPU-saving win is worth the
diff --git a/docs/security/AGENTROUTER_WAF.md b/docs/security/AGENTROUTER_WAF.md
index 8aaaa941c6..0311da0a07 100644
--- a/docs/security/AGENTROUTER_WAF.md
+++ b/docs/security/AGENTROUTER_WAF.md
@@ -1,5 +1,5 @@
 ---
-title: "AgentRouter WAF"
+title: "agentrouter.org WAF (Web Application Firewall)"
 version: 3.8.50
 lastUpdated: 2026-08-03
 ---
@@ -94,4 +94,4 @@ The current filter is overly aggressive — it blocks "Lorem ipsum" in
 `tool_result` blocks even though the operator clearly did not intend to
 inject a prompt. Operators who want this fixed at the source should
 contact `agentrouter.org` to report the false positives. The blocklist
-above is the empirical result of probing the upstream as of 2026-08-03.
\ No newline at end of file
+above is the empirical result of probing the upstream as of 2026-08-03.
diff --git a/open-sse/handlers/chatCore.ts b/open-sse/handlers/chatCore.ts
index 99be5cda82..fe9da81aeb 100644
--- a/open-sse/handlers/chatCore.ts
+++ b/open-sse/handlers/chatCore.ts
@@ -302,6 +302,7 @@ import {
   updateFromResponseBody,
   initializeRateLimits,
 } from "../services/rateLimitManager.ts";
+import * as localLimiterErrors from "../services/rateLimitManager/errors.ts";
 import {
   acquire as acquireAccountSemaphore,
   markBlocked as markAccountSemaphoreBlocked,
@@ -3362,27 +3363,22 @@ export async function handleChatCore({
         errorCode: error.code,
       };
     }
-    // abort(reason) can reject the upstream fetch with a raw string reason
-    // (e.g. "request_signal_aborted") that has no `name`/`status`; classify
-    // via isLocalStreamLifecycleError so those map to 499 instead of falling
-    // through to the 502 provider-failure default.
+    // abort(reason) can reject with a raw string lacking `name`/`status`; classify
+    // it through isLocalStreamLifecycleError so it maps to 499 rather than the
+    // 502 provider-failure default.
     const isRequestAborted = isLocalStreamLifecycleError(error);
-    // #8376: an unreachable upstream proxy (ECONNREFUSED/ECONNRESET/...) is tagged by
-    // proxyFetch.ts (tagProxyUnreachable) with `.errorCode = "proxy_unreachable"` before
-    // it reaches this catch. Classify it explicitly to 502 instead of falling through
-    // the generic `error.status` branch (a raw connect-refused error has no `.status` at
-    // all, so it used to collapse into an ordinary 502/504 the provider-breaker predicate
-    // can't tell apart from a per-model 5xx).
+    // #8376: proxyFetch tags unreachable transport failures so they remain
+    // distinguishable from ordinary provider 5xx responses.
     const isProxyUnreachableFailure =
       !isRequestAborted && (error as { errorCode?: unknown })?.errorCode === "proxy_unreachable";
     const errorCode = getUpstreamErrorIdentifier(error);
-    const isLocalQueueTimeout = errorCode === "RATE_LIMIT_QUEUE_TIMEOUT";
+    const localRateLimitFailure = localLimiterErrors.getClientSafeLocalRateLimitError(error);
     const failureStatus = isRequestAborted
       ? 499
       : isProxyUnreachableFailure
         ? HTTP_STATUS.BAD_GATEWAY
-        : isLocalQueueTimeout
-          ? HTTP_STATUS.SERVICE_UNAVAILABLE
+        : localRateLimitFailure
+          ? localRateLimitFailure.status
           : error.name === "TimeoutError" || error.name === "BodyTimeoutError"
             ? HTTP_STATUS.GATEWAY_TIMEOUT
             : error.status && typeof error.status === "number"
@@ -3390,8 +3386,9 @@ export async function handleChatCore({
               : HTTP_STATUS.BAD_GATEWAY;
     const failureMessage = isRequestAborted
       ? "Request aborted"
-      : formatProviderError(error, provider, model, failureStatus);
-    const upstreamErrorCode = isProxyUnreachableFailure ? "proxy_unreachable" : errorCode;
+      : formatProviderError(localRateLimitFailure ?? error, provider, model, failureStatus);
+    const upstreamErrorCode =
+      localRateLimitFailure?.code ?? (isProxyUnreachableFailure ? "proxy_unreachable" : errorCode);
     // Tag our own deadline timeouts (fetch-start TimeoutError / body BodyTimeoutError,
     // both surfaced as a 504) as "upstream_timeout" so the cooldown layer can tell a
     // slow-but-not-failed request apart from a real provider 5xx. (Antigravity already
@@ -3441,19 +3438,22 @@ export async function handleChatCore({
         upstreamErrorCode,
         upstreamErrorType
       );
+      localLimiterErrors.markTrustedLocalRateLimitResponse(result.response, error);
       return {
         ...result,
         errorType: upstreamErrorType,
         errorCode: upstreamErrorCode,
       };
     }
-    return createErrorResult(
+    const result = createErrorResult(
       failureStatus,
       failureMessage,
       null,
       upstreamErrorCode,
       upstreamErrorType
     );
+    localLimiterErrors.markTrustedLocalRateLimitResponse(result.response, error);
+    return result;
   }
   let upstreamErrorParsed = false;
   let parsedStatusCode = providerResponse.status;
diff --git a/open-sse/services/combo.ts b/open-sse/services/combo.ts
index 7db96ae231..7b73108df5 100644
--- a/open-sse/services/combo.ts
+++ b/open-sse/services/combo.ts
@@ -1594,7 +1594,7 @@ export async function handleComboChat({
                       : undefined,
                 }
               : undefined;
-          const scopedFailure = isScopedFailure(result.status, errorText, structuredError);
+          const scopedFailure = isScopedFailure(result, errorText, structuredError);
 
           // #8375: input-bound request-scoped failures (context_length_exceeded) are
           // deterministic for the same input — retrying on other accounts of the same
@@ -1675,6 +1675,7 @@ export async function handleComboChat({
             rawModel,
             isTokenLimitBreach,
             allAccountsRateLimited: false,
+            requestScopedFailure: scopedFailure,
             sets: { exhaustedProviders, exhaustedConnections, transientRateLimitedProviders },
             log,
             tag: "COMBO",
@@ -1767,6 +1768,7 @@ export async function handleComboChat({
           const isTransient =
             !isStreamReadinessFailure &&
             !isTokenLimitBreach &&
+            !scopedFailure &&
             [408, 429, 500, 502, 503, 504].includes(result.status);
           if (retry < maxRetries && isTransient && !providerExhausted) {
             if (
@@ -2841,7 +2843,7 @@ async function handleRoundRobinCombo({
                     : undefined,
               }
             : undefined;
-        const scopedFailure = isScopedFailure(result.status, errorText, structuredError);
+        const scopedFailure = isScopedFailure(result, errorText, structuredError);
         const fallbackResult = checkFallbackError(
           result.status,
           errorText,
@@ -2880,6 +2882,7 @@ async function handleRoundRobinCombo({
           rawModel: parseModel(modelStr).model || modelStr,
           isTokenLimitBreach,
           allAccountsRateLimited: isAllAccountsRateLimited,
+          requestScopedFailure: scopedFailure,
           sets: { exhaustedProviders, exhaustedConnections, transientRateLimitedProviders },
           log,
           tag: "COMBO-RR",
@@ -2913,6 +2916,7 @@ async function handleRoundRobinCombo({
         const isTransient =
           !isStreamReadinessFailure &&
           !isTokenLimitBreach &&
+          !scopedFailure &&
           [408, 429, 500, 502, 503, 504].includes(result.status);
         if (retry < maxRetries && isTransient && !providerExhausted) {
           continue;
diff --git a/open-sse/services/combo/comboPredicates.ts b/open-sse/services/combo/comboPredicates.ts
index cc29f710fa..93f1568c96 100644
--- a/open-sse/services/combo/comboPredicates.ts
+++ b/open-sse/services/combo/comboPredicates.ts
@@ -12,6 +12,7 @@ import { isSelfInflictedUpstreamTimeout } from "../../handlers/chatCore/cooldown
 import { isLocalStreamLifecycleError } from "@/shared/utils/circuitBreaker";
 import { CONTEXT_OVERFLOW_PATTERNS, MODEL_ACCESS_DENIED_PATTERNS } from "../accountFallback.ts";
 import { isResourceNotFoundResponse } from "../errorClassifier.ts";
+import { getTrustedLocalRateLimitResponse } from "../rateLimitManager/errors.ts";
 import type { ResolvedComboTarget } from "./types.ts";
 
 // Status codes that should mark round-robin target semaphores as cooling down.
@@ -190,13 +191,13 @@ export function shouldRecordProviderBreakerFailure(args: {
   );
 }
 
-const REQUEST_SCOPED_UPSTREAM_ERROR_CODES = new Set([
-  "context_length_exceeded",
-  "upstream_empty_response",
-  "upstream_response_failed",
+const REQUEST_SCOPED_UPSTREAM_ERROR_CODES: Record<string, true> = {
+  context_length_exceeded: true,
+  upstream_empty_response: true,
+  upstream_response_failed: true,
   // Local combo per-target timer (targetTimeoutRunner) — not a connection health signal.
-  "combo_target_timeout",
-]);
+  combo_target_timeout: true,
+};
 
 /** Request/model-specific failures must not poison provider-wide resilience state. */
 export function isRequestScopedUpstreamFailure(error?: {
@@ -205,18 +206,19 @@ export function isRequestScopedUpstreamFailure(error?: {
 }): boolean {
   const code = typeof error?.code === "string" ? error.code.toLowerCase() : "";
   const type = typeof error?.type === "string" ? error.type.toLowerCase() : "";
-  return REQUEST_SCOPED_UPSTREAM_ERROR_CODES.has(code) || type === "context_length_exceeded";
+  return REQUEST_SCOPED_UPSTREAM_ERROR_CODES[code] === true || type === "context_length_exceeded";
 }
 
 /** Request-scoped classification that also has access to the HTTP body. */
 export function isComboRequestScopedFailure(
-  status: number,
+  response: Response,
   errorText: string,
   error?: { code?: string | null; type?: string | null }
 ): boolean {
   return (
+    getTrustedLocalRateLimitResponse(response) !== null ||
     isRequestScopedUpstreamFailure(error) ||
-    (status === 404 && isResourceNotFoundResponse(errorText))
+    (response.status === 404 && isResourceNotFoundResponse(errorText))
   );
 }
 
@@ -255,6 +257,7 @@ export function isInputBoundRequestFailure(error?: {
 export function shouldSkipConnDisable(
   result: {
     status: number;
+    response?: Response;
     errorCode?: string | null;
     errorType?: string | null;
     error?: unknown;
@@ -270,6 +273,7 @@ export function shouldSkipConnDisable(
     // Client abort surfaced as a bare error (no statusCode → defaults to 502):
     // a local lifecycle event, not a provider failure (#4602 policy).
     isLocalStreamLifecycleError(result.error) ||
+    (result.response ? getTrustedLocalRateLimitResponse(result.response) !== null : false) ||
     result.errorCode === "plugin_block" ||
     result.errorType === "plugin_block" ||
     (is401 && hasExtraKeys) ||
diff --git a/open-sse/services/combo/targetExhaustion.ts b/open-sse/services/combo/targetExhaustion.ts
index 02989b68f2..c4a880e79a 100644
--- a/open-sse/services/combo/targetExhaustion.ts
+++ b/open-sse/services/combo/targetExhaustion.ts
@@ -65,6 +65,7 @@ export type ApplyComboTargetExhaustionOptions = {
   rawModel: string;
   isTokenLimitBreach: boolean;
   allAccountsRateLimited: boolean;
+  requestScopedFailure: boolean;
   sets: ComboExhaustionSets;
   log: ComboLogger;
   tag: string;
@@ -154,12 +155,25 @@ function isProviderQuotaExhausted(
   provider: string | null | undefined,
   opts: Pick<
     ApplyComboTargetExhaustionOptions,
-    "rawModel" | "fallbackResult" | "structuredError" | "errorText" | "allAccountsRateLimited"
+    | "rawModel"
+    | "fallbackResult"
+    | "structuredError"
+    | "errorText"
+    | "allAccountsRateLimited"
+    | "requestScopedFailure"
   >
 ): boolean {
-  const { rawModel, fallbackResult, structuredError, errorText, allAccountsRateLimited } = opts;
+  const {
+    rawModel,
+    fallbackResult,
+    structuredError,
+    errorText,
+    allAccountsRateLimited,
+    requestScopedFailure,
+  } = opts;
   return (
     Boolean(provider && provider !== "unknown") &&
+    !(requestScopedFailure || isRequestScopedUpstreamFailure(structuredError)) &&
     !hasPerModelQuota(provider as string, rawModel) &&
     (isProviderExhaustedReason(fallbackResult) ||
       classifyErrorText(structuredError?.code || errorText) === RateLimitReason.QUOTA_EXHAUSTED ||
@@ -189,7 +203,17 @@ function markTransientOrConnectionLevel(
   target: ResolvedComboTarget,
   opts: ApplyComboTargetExhaustionOptions
 ): void {
-  const { result, errorText, rawModel, isTokenLimitBreach, sets, log, tag, structuredError } = opts;
+  const {
+    result,
+    errorText,
+    rawModel,
+    isTokenLimitBreach,
+    requestScopedFailure,
+    sets,
+    log,
+    tag,
+    structuredError,
+  } = opts;
   const provider = target.provider;
   if (result.status === 429 && !isTokenLimitBreach && provider && provider !== "unknown") {
     sets.transientRateLimitedProviders.add(provider);
@@ -201,6 +225,7 @@ function markTransientOrConnectionLevel(
     log,
     tag,
     rawModel,
+    requestScopedFailure,
     structuredError,
   });
 }
@@ -244,16 +269,25 @@ function markConnectionLevelExhaustion(
   target: ResolvedComboTarget,
   opts: Pick<
     ApplyComboTargetExhaustionOptions,
-    "result" | "errorText" | "sets" | "log" | "tag" | "rawModel" | "structuredError"
+    | "result"
+    | "errorText"
+    | "sets"
+    | "log"
+    | "tag"
+    | "rawModel"
+    | "requestScopedFailure"
+    | "structuredError"
   >
 ): void {
-  const { result, errorText, sets, log, tag, rawModel, structuredError } = opts;
+  const { result, errorText, sets, log, tag, rawModel, requestScopedFailure, structuredError } =
+    opts;
   const provider = target.provider;
   if (
     !provider ||
     provider === "unknown" ||
     !CONNECTION_LEVEL_ERROR_STATUSES.includes(result.status) ||
     isProviderCircuitOpenResult(result, errorText) ||
+    requestScopedFailure ||
     isRequestScopedUpstreamFailure(structuredError) ||
     // #5085: empty-content 502 is a healthy connection returning no body — model-level, not
     // connection-level. Don't exhaust the provider; let the remaining legs (incl. same-provider)
diff --git a/open-sse/services/rateLimitManager.ts b/open-sse/services/rateLimitManager.ts
index cb4350b2f8..fbc7d0d2fc 100644
--- a/open-sse/services/rateLimitManager.ts
+++ b/open-sse/services/rateLimitManager.ts
@@ -13,12 +13,7 @@ import { parseRetryAfterFromBody } from "./accountFallback.ts";
 import { getAntigravityQuotaFamily } from "./antigravityQuotaFamily.ts";
 import { getProviderCategory } from "../config/providerRegistry.ts";
 import { getCodexRateLimitKey } from "../executors/codex.ts";
-import {
-  getProviderDefaultRateLimit,
-  setProviderQuotaOverrides,
-} from "./providerDefaultRateLimit.ts";
-import { keyContainsConnection, RollingRpmGate } from "./rollingRpmGate.ts";
-import { toNumber } from "@/shared/utils/numeric";
+import { awaitProviderDefaultSlot, setProviderQuotaOverrides } from "./providerDefaultRateLimit.ts";
 import {
   DEFAULT_RESILIENCE_SETTINGS,
   resolveResilienceSettings,
@@ -31,6 +26,13 @@ import {
   toPlainHeaders,
 } from "./rateLimitManager/headers";
 import { checkQueueAdmission } from "./rateLimitManager/admission";
+import {
+  markLocalRateLimitError,
+  RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+  RATE_LIMIT_QUEUE_WEDGED_CODE,
+} from "./rateLimitManager/errors";
+import { LimiterWedgeWatchdog, WATCHDOG_INTERVAL_MS } from "./rateLimitManager/wedgeWatchdog";
+import { toNumber } from "@/shared/utils/numeric";
 
 interface LearnedLimitEntry {
   provider: string;
@@ -44,38 +46,17 @@ interface LearnedLimitEntry {
 interface LimiterUpdateSettings {
   maxConcurrent?: number | null;
   minTime: number;
+  reservoir?: number | null;
+  reservoirRefreshAmount?: number | null;
+  reservoirRefreshInterval?: number | null;
 }
 
 type JsonRecord = Record<string, unknown>;
-type QueueTimeoutReason = "local-queue" | "upstream-cooldown";
 
 function toRecord(value: unknown): JsonRecord {
   return value && typeof value === "object" && !Array.isArray(value) ? (value as JsonRecord) : {};
 }
 
-function createQueueTimeoutError(
-  provider: string,
-  model: string | null,
-  maxWaitMs: number,
-  reason: QueueTimeoutReason = "local-queue",
-  cause?: unknown
-) {
-  const target = model ? `${provider}/${model}` : provider;
-  const message =
-    reason === "upstream-cooldown"
-      ? `Request dropped after waiting ${maxWaitMs}ms for an upstream rate-limit cooldown for ${target}. ` +
-        `The provider cooldown outlasted OmniRoute's local wait budget; this is not local queue saturation.`
-      : `Request dropped after exceeding the local rate-limit queue budget maxWaitMs (${maxWaitMs}ms) for ` +
-        `${target} — this is OmniRoute's request queue ` +
-        `(resilienceSettings.requestQueue.maxWaitMs), not an upstream timeout. Raise it in ` +
-        `Settings → Resilience if this is queue saturation rather than a slow provider.`;
-  const queueErr = new Error(message, cause === undefined ? undefined : { cause }) as Error & {
-    code?: string;
-  };
-  queueErr.code = "RATE_LIMIT_QUEUE_TIMEOUT";
-  return queueErr;
-}
-
 function isNodeTestRunnerChild(): boolean {
   return typeof process.env.NODE_TEST_CONTEXT === "string";
 }
@@ -105,7 +86,6 @@ const connectionRateLimitOverrides = new Map<string, Record<string, number>>();
 // Store learned limits for persistence (debounced)
 const learnedLimits: Record<string, LearnedLimitEntry> = {};
 const MAX_LEARNED_LIMITS = 200;
-const INACTIVE_LIMITER_MS = 10 * 60 * 1000;
 const limiterLastUsed = new Map<string, number>();
 let persistTimer: ReturnType<typeof setTimeout> | null = null;
 const pendingAsyncOperations = new Set<Promise<unknown>>();
@@ -116,17 +96,24 @@ let initialized = false;
 
 let currentRequestQueueSettings: RequestQueueSettings = DEFAULT_RESILIENCE_SETTINGS.requestQueue;
 
-// Watchdog: detect Bottleneck limiters that are wedged (queue has work, but no
-// jobs are dispatched). RPM admission happens before Bottleneck, so a queued
-// Bottleneck job with no active work is a concurrency scheduler failure.
-const lastDispatchAt = new Map<string, number>();
-let nextJobTraceId = 1;
+const limiterEffectiveSettings = new WeakMap<Bottleneck, Bottleneck.ConstructorOptions>();
+const preservedReplacementSettings = new Map<string, Bottleneck.ConstructorOptions>();
+const limiterWatchdog = new LimiterWedgeWatchdog({
+  limiters,
+  limiterLastUsed,
+  limiterEffectiveSettings,
+  preservedReplacementSettings,
+  trackBackground: (promise) => {
+    trackAsyncOperation(promise);
+  },
+  log: logRateLimit,
+  warn: warnRateLimit,
+});
 let watchdogInterval: ReturnType<typeof setInterval> | null = null;
-const WATCHDOG_INTERVAL_MS = 30_000;
-// Threshold has to exceed any legitimate gap caused by adaptive minTime while
-// still catching the actual wedge case we observed (queue stalled for 3+
-// minutes with no progress).
-const WEDGE_THRESHOLD_MS = 120_000;
+
+type LimiterFactory = (options: Bottleneck.ConstructorOptions) => Bottleneck;
+const defaultLimiterFactory: LimiterFactory = (options) => new Bottleneck(options);
+let limiterFactory: LimiterFactory = defaultLimiterFactory;
 
 /**
  * Env-var override for the auto-enable safety net. Highest priority — wins
@@ -143,10 +130,19 @@ function isAutoEnableActive(settings: RequestQueueSettings): boolean {
   return settings.autoEnableApiKeyProviders;
 }
 
-// Bottleneck handles concurrency and pacing. RPM is enforced by the rolling
-// lease limiter above rather than by a fixed-window reservoir.
+// Sentinels for "no rate limit" / effectively infinite capacity. The reservoir
+// value uses Number.MAX_SAFE_INTEGER so the bucket can never realistically be
+// exhausted; maxConcurrent uses a smaller-but-still-vast ceiling since
+// Bottleneck tracks concurrent jobs in memory and an unbounded number would
+// risk internal counter overflow under sustained pressure.
+const EFFECTIVELY_INFINITE = Number.MAX_SAFE_INTEGER;
 const EFFECTIVELY_INFINITE_CONCURRENCY = 1000;
 
+// Resolve an RPM override. 0 or missing means "infinite" (no rate cap).
+function resolveRpm(override: number | undefined | null): number {
+  return typeof override === "number" && override > 0 ? override : EFFECTIVELY_INFINITE;
+}
+
 // Resolve a minTime override. 0 or missing means "no minimum gap".
 function resolveMinTime(override: number | undefined | null): number {
   return typeof override === "number" && override > 0 ? override : 0;
@@ -158,62 +154,38 @@ function resolveMaxConcurrent(override: number | undefined | null): number {
 }
 
 function buildLimiterDefaults() {
+  // 0 or missing values mean "infinite" / no rate limit applies. This treats
+  // the global request-queue settings the same way per-connection overrides
+  // are interpreted (see resolveRpm / resolveMinTime / resolveMaxConcurrent).
   return {
     maxConcurrent: resolveMaxConcurrent(currentRequestQueueSettings.concurrentRequests),
     minTime: resolveMinTime(currentRequestQueueSettings.minTimeBetweenRequestsMs),
+    reservoir: resolveRpm(currentRequestQueueSettings.requestsPerMinute),
+    reservoirRefreshAmount: resolveRpm(currentRequestQueueSettings.requestsPerMinute),
+    reservoirRefreshInterval: 60 * 1000,
   };
 }
 
-/**
- * Apply new settings to a Bottleneck limiter and re-arm its reservoir-refresh
- * heartbeat.
- *
- * Bottleneck 2.19.5 (frozen upstream dependency, no release since 2019) has a
- * bug in `LocalDatastore#_startHeartbeat()`
- * (node_modules/bottleneck/lib/LocalDatastore.js:29,56): the guard
- * `if (this.heartbeat == null && ...)` only (re)creates the periodic
- * reservoir-refresh interval the FIRST time it runs. Every later call —
- * including the one `updateSettings()` itself triggers internally — falls
- * into the `else` branch and does `clearInterval(this.heartbeat)` WITHOUT
- * resetting `this.heartbeat` back to `null`. Because the stale reference is
- * left in place, every future `_startHeartbeat()` call keeps taking the same
- * dead `else` branch: the periodic reservoir refresh is gone forever after
- * the FIRST manual `updateSettings()` call on a limiter — every limiter here
- * starts with a live heartbeat (buildLimiterDefaults() always sets
- * reservoirRefreshInterval/reservoirRefreshAmount), so that "first call" is
- * whichever of the 5 updateSettings() call sites in this file runs first.
- *
- * Work around it here instead of patching node_modules: null out the stale
- * reference ourselves and re-invoke `_startHeartbeat()` so it takes the
- * "start a fresh interval" branch again. Every `limiter.updateSettings(...)`
- * call in this file MUST go through this helper, never Bottleneck's method
- * directly.
- */
-async function applyLimiterSettings(
+function updateLimiterSettings(
   limiter: Bottleneck,
   updates: Bottleneck.ConstructorOptions
-): Promise<void> {
-  await limiter.updateSettings(updates);
-  const store = (
-    limiter as unknown as {
-      _store?: {
-        heartbeat?: ReturnType<typeof setInterval> | null;
-        _startHeartbeat?: () => void;
-      };
-    }
-  )._store;
-  if (store && typeof store._startHeartbeat === "function") {
-    if (store.heartbeat != null) clearInterval(store.heartbeat);
-    store.heartbeat = null;
-    store._startHeartbeat();
+): Bottleneck {
+  const effective = limiterEffectiveSettings.get(limiter) ?? {};
+  limiterEffectiveSettings.set(limiter, { ...effective, ...updates });
+  return limiter.updateSettings(updates);
+}
+
+function updateAllLimiterSettings() {
+  const defaults = buildLimiterDefaults();
+  for (const limiter of limiters.values()) {
+    updateLimiterSettings(limiter, defaults);
   }
 }
 
-async function updateAllLimiterSettings() {
-  const defaults = buildLimiterDefaults();
-  await Promise.all(
-    Array.from(limiters.values(), (limiter) => applyLimiterSettings(limiter, defaults))
-  );
+function clearPreservedReplacementSettings(connectionId: string): void {
+  for (const key of preservedReplacementSettings.keys()) {
+    if (key.includes(connectionId)) preservedReplacementSettings.delete(key);
+  }
 }
 
 function reconcileEnabledConnections(
@@ -246,9 +218,8 @@ function reconcileEnabledConnections(
       nextEnabledConnections.add(connectionId);
       autoCount++;
 
-      // Route through getLimiter so the `queued`/`executing` listeners and
-      // lastDispatchAt heartbeat are wired up — otherwise the watchdog sees
-      // `stalledMs = now - 0` and falsely flags healthy idle limiters as wedged.
+      // Route through getLimiter so the queue-progress listeners are wired up.
+      // Otherwise a limiter created here could not be evaluated safely by the watchdog.
       getLimiter(provider, connectionId);
     }
   }
@@ -269,82 +240,16 @@ function reconcileEnabledConnections(
   };
 }
 
-function watchdogTick() {
-  const now = Date.now();
-  rpmGate.cleanupExpired(now);
-  // Clean up idle limiters that haven't been used recently
-  for (const [key, limiter] of Array.from(limiters)) {
-    const lastUsed = limiterLastUsed.get(key) ?? 0;
-    if (now - lastUsed > INACTIVE_LIMITER_MS) {
-      const counts = limiter.counts();
-      if (
-        counts.RECEIVED === 0 &&
-        counts.QUEUED === 0 &&
-        counts.RUNNING === 0 &&
-        counts.EXECUTING === 0
-      ) {
-        limiters.delete(key);
-        lastDispatchAt.delete(key);
-        limiterLastUsed.delete(key);
-        logRateLimit(
-          `🧹 [RATE-LIMIT] Evicting idle limiter: ${key} (inactive for ${Math.round((now - lastUsed) / 1000)}s)`
-        );
-        trackAsyncOperation(limiter.disconnect());
-      }
-    }
-  }
-  for (const [key, limiter] of Array.from(limiters)) {
-    const counts = limiter.counts();
-    // RECEIVED-only work is still active and must not be evicted. Once a job
-    // is stably queued, Bottleneck reports it in QUEUED with RECEIVED=0; that
-    // is the state the wedge detector is designed to recover.
-    if (counts.RECEIVED > 0 || counts.QUEUED === 0) continue;
-    if (counts.RUNNING > 0 || counts.EXECUTING > 0) continue;
-    const lastDispatch = lastDispatchAt.get(key);
-    // No heartbeat yet → seed it and skip this tick. Prevents false wedge
-    // detection on a brand-new limiter or one created outside getLimiter.
-    if (lastDispatch === undefined) {
-      lastDispatchAt.set(key, now);
-      continue;
-    }
-    const stalledMs = now - lastDispatch;
-    if (stalledMs < WEDGE_THRESHOLD_MS) continue;
-
-    warnRateLimit(
-      `🚨 [RATE-LIMIT] WEDGED: ${key} received=${counts.RECEIVED} queued=${counts.QUEUED} running=0 executing=0 stalled=${stalledMs}ms — force-resetting`
-    );
-    // Live incident (log id 1784465227489-a2cbc0): disconnect() releases the
-    // heartbeat timer but does NOT reject the QUEUED jobs already sitting on
-    // this instance — withRateLimit's `limiter.schedule()` for those callers
-    // then just hangs forever (nothing will ever dequeue them; getLimiter()
-    // only hands out a FRESH instance to future callers), leaving the
-    // dispatch orphaned until the outer ~300s per-target timeout eventually
-    // aborts it. Real clients routinely give up (and retry) well before that
-    // — this specific incident's client aborted at ~60s having never reached
-    // the provider at all (queued=2 running=0 executing=0 the entire time).
-    //
-    // stop({ dropWaitingJobs: true }) rejects exactly the RECEIVED/QUEUED/
-    // RUNNING jobs on THIS instance immediately (Bottleneck's own contract —
-    // see node_modules/bottleneck/bottleneck.d.ts StopOptions) so those
-    // withRateLimit() callers reject right away instead of hanging, letting
-    // combo's fallback/cooldown-wait engage within seconds instead of minutes.
-    // This is safe against the previously-documented "spurious 502 bursts"
-    // concern: the wedge condition checked above already requires
-    // RUNNING === 0 && EXECUTING === 0, so no job that's actually progressing
-    // can be caught by this — only ones already confirmed stuck. The instance
-    // is deleted from `limiters` synchronously (above) before this call, so
-    // no future getLimiter() call can ever hand out this now-stopped instance
-    // — the "permanently rejects future .schedule()" behavior stop() has is
-    // therefore moot; nothing will call .schedule() on it again.
-    evictWedgeLimiter(key, limiter);
-  }
-}
-
 let shutdownHandlersRegistered = false;
 
 export function startRateLimitWatchdog(): void {
   if (watchdogInterval) return;
-  watchdogInterval = setInterval(watchdogTick, WATCHDOG_INTERVAL_MS);
+  watchdogInterval = setInterval(() => {
+    const run = trackAsyncOperation(limiterWatchdog.run());
+    void run.then(undefined, (error) => {
+      errorRateLimit("[RATE-LIMIT] Watchdog scan failed:", error);
+    });
+  }, WATCHDOG_INTERVAL_MS);
   watchdogInterval.unref?.();
   // Register SIGTERM/SIGINT shutdown handlers once, lazily, on first watchdog start.
   // Registering here (rather than at module load) avoids interfering with test runner
@@ -362,54 +267,18 @@ export function stopRateLimitWatchdog(): void {
   watchdogInterval = null;
 }
 
-export function __installLimiterForTests(
-  provider: string,
-  connectionId: string,
-  limiter: Bottleneck,
-  model = null
-): void {
-  const key = getLimiterKey(provider, connectionId, model);
-  limiters.set(key, limiter);
-  lastDispatchAt.set(key, Date.now());
-  limiterLastUsed.set(key, Date.now());
-}
-
-export function __runRateLimitWatchdogForTests(): void {
-  watchdogTick();
-}
-
-export function __getLimiterForTests(provider: string, connectionId: string, model = null) {
-  return getLimiter(provider, connectionId, model);
-}
-
-export function __setLastDispatchAtForTests(
-  provider: string,
-  connectionId: string,
-  model: string | null,
-  timestamp: number
-): void {
-  lastDispatchAt.set(getLimiterKey(provider, connectionId, model), timestamp);
-}
-
-function evictWedgeLimiter(key: string, limiter: Bottleneck): void {
-  if (limiters.get(key) !== limiter) return;
-  evictLimiterAndDropQueued(key, limiter, "rate-limit-watchdog-wedge-reset");
-}
-
 /**
  * Gracefully stop all limiters for process shutdown.
- * ONLY call this from SIGTERM/SIGINT handlers — not during runtime resets.
- * Calling .stop() during runtime (e.g. on 429 or connection disable) permanently
- * rejects future .schedule() calls, causing 502 bursts. This function is the
- * sole legitimate use of limiter.stop() in this module.
+ * Runtime wedge recovery also uses stop(), but only after synchronously
+ * removing that limiter from the cache so it can never accept new work.
  */
 function shutdownLimiters(): void {
   for (const limiter of limiters.values()) {
     limiter.stop({ dropWaitingJobs: false });
   }
   limiters.clear();
-  lastDispatchAt.clear();
   limiterLastUsed.clear();
+  preservedReplacementSettings.clear();
 }
 
 // Only register shutdown handlers when there are active limiters to shut down.
@@ -454,10 +323,13 @@ export async function initializeRateLimits() {
     // budget + concurrency cap (nvidia today). No-op for every provider without
     // an entry in either providerQuotaOverrides or PROVIDER_DEFAULT_RATE_LIMITS.
     setProviderQuotaOverrides(resilience.providerQuotaOverrides);
+    const { explicitCount, autoCount } = reconcileEnabledConnections(
+      connections as unknown[],
+      currentRequestQueueSettings
+    );
+    updateAllLimiterSettings();
 
-    // Load per-connection rate limit overrides before reconciliation can create
-    // any limiter. The RPM gate reads these overrides at admission time, and
-    // Bottleneck still needs the non-RPM connection settings immediately.
+    // Load per-connection rate limit overrides
     connectionRateLimitOverrides.clear();
     for (const conn of connections as Array<Record<string, unknown>>) {
       const overrides = conn.rateLimitOverrides;
@@ -466,12 +338,6 @@ export async function initializeRateLimits() {
       }
     }
 
-    const { explicitCount, autoCount } = reconcileEnabledConnections(
-      connections as unknown[],
-      currentRequestQueueSettings
-    );
-    updateAllLimiterSettings();
-
     if (explicitCount > 0 || autoCount > 0) {
       logRateLimit(
         `🛡️ [RATE-LIMIT] Loaded ${explicitCount} explicit + ${autoCount} auto-enabled protection(s)`
@@ -491,16 +357,21 @@ export async function initializeRateLimits() {
 
 export async function applyRequestQueueSettings(nextSettings: RequestQueueSettings) {
   currentRequestQueueSettings = { ...nextSettings };
+  // Global policy changes invalidate snapshots from the previous generation.
+  preservedReplacementSettings.clear();
   const { getCachedProviderConnections } = await import("@/lib/localDb");
   const connections = await getCachedProviderConnections();
+  // Also discard any snapshot created while the asynchronous DB read yielded.
+  preservedReplacementSettings.clear();
   reconcileEnabledConnections(connections as unknown[], currentRequestQueueSettings);
-  await updateAllLimiterSettings();
+  updateAllLimiterSettings();
 }
 
 /**
  * Get or create a limiter for a given provider+connection combination
  */
 export function enableRateLimitProtection(connectionId) {
+  if (!enabledConnections.has(connectionId)) clearPreservedReplacementSettings(connectionId);
   enabledConnections.add(connectionId);
 }
 
@@ -509,14 +380,19 @@ export function enableRateLimitProtection(connectionId) {
  */
 export function disableRateLimitProtection(connectionId) {
   enabledConnections.delete(connectionId);
-  // Drop queued jobs before evicting the limiter. Otherwise disconnect() leaves
-  // callers waiting on an instance that is no longer reachable from the cache.
+  clearPreservedReplacementSettings(connectionId);
+  // Ordinary administrative eviction uses disconnect(), not stop(), so
+  // in-flight requests can finish. Wedge recovery is the deliberate exception:
+  // it removes the limiter from the cache first, then stops it to settle jobs
+  // that were already proven stranded.
   for (const [key, limiter] of Array.from(limiters)) {
-    if (keyContainsConnection(key, connectionId)) {
-      evictLimiterAndDropQueued(key, limiter, "rate-limit-connection-disabled");
+    if (key.includes(connectionId)) {
+      limiters.delete(key);
+      limiterWatchdog.forget(limiter);
+      limiterLastUsed.delete(key);
+      trackAsyncOperation(limiter.disconnect());
     }
   }
-  rpmGate.clearConnection(connectionId);
 }
 
 /**
@@ -542,13 +418,16 @@ export function refreshConnectionRateLimits(connectionId, overrides) {
   } else {
     connectionRateLimitOverrides.set(connectionId, overrides);
   }
+  clearPreservedReplacementSettings(connectionId);
   // Evict limiters referencing this connection so they get recreated on next use
   for (const [key, limiter] of Array.from(limiters)) {
-    if (keyContainsConnection(key, connectionId)) {
-      evictLimiterAndDropQueued(key, limiter, "rate-limit-settings-refresh");
+    if (key.includes(connectionId)) {
+      limiters.delete(key);
+      limiterWatchdog.forget(limiter);
+      limiterLastUsed.delete(key);
+      trackAsyncOperation(limiter.disconnect());
     }
   }
-  rpmGate.clearConnection(connectionId);
 }
 
 /**
@@ -571,46 +450,51 @@ function getLimiterKey(provider, connectionId, model = null) {
   return `${provider}:${connectionId}`;
 }
 
-const rpmGate = new RollingRpmGate({
-  getGlobalRpm: () => currentRequestQueueSettings.requestsPerMinute,
-  getProviderWindow: getProviderDefaultRateLimit,
-  getConnectionRpm: (connectionId) => connectionRateLimitOverrides.get(connectionId)?.rpm,
-  getLimiterKey,
-  createQueueTimeoutError: (provider, model, maxWaitMs, reason) =>
-    createQueueTimeoutError(provider, model, maxWaitMs, reason),
-});
-
 function getLimiter(provider, connectionId, model = null) {
   const key = getLimiterKey(provider, connectionId, model);
 
   if (!limiters.has(key)) {
-    const defaults = buildLimiterDefaults();
-    const overrides = connectionRateLimitOverrides.get(connectionId);
-    if (overrides) {
-      // 0 (or missing) means "no override — fall through to buildLimiterDefaults()".
-      if (typeof overrides.maxConcurrent === "number" && overrides.maxConcurrent > 0) {
-        defaults.maxConcurrent = overrides.maxConcurrent;
+    const preserved = preservedReplacementSettings.get(key);
+    let options: Bottleneck.ConstructorOptions;
+    if (preserved) {
+      preservedReplacementSettings.delete(key);
+      options = { ...preserved, id: key };
+    } else {
+      const defaults = buildLimiterDefaults();
+      const overrides = connectionRateLimitOverrides.get(connectionId);
+      if (overrides) {
+        // 0 (or missing) means "no override — fall through to buildLimiterDefaults()".
+        // Without this guard, an rpm of 0 sets reservoir=0, which Bottleneck treats
+        // as depleted and blocks all requests indefinitely.
+        if (typeof overrides.maxConcurrent === "number" && overrides.maxConcurrent > 0) {
+          defaults.maxConcurrent = overrides.maxConcurrent;
+        }
+        if (typeof overrides.minTime === "number" && overrides.minTime > 0) {
+          defaults.minTime = overrides.minTime;
+        }
+        if (typeof overrides.rpm === "number" && overrides.rpm > 0) {
+          defaults.reservoir = overrides.rpm;
+          defaults.reservoirRefreshAmount = overrides.rpm;
+          defaults.reservoirRefreshInterval = 60 * 1000;
+        }
+        // TODO: TPM/TPD integration requires separate token and request buckets.
       }
-      if (typeof overrides.minTime === "number" && overrides.minTime > 0) {
-        defaults.minTime = overrides.minTime;
-      }
-      // TODO: TPM/TPD integration — requires a token-bucket vs request-bucket
-      // separation. RPM is handled by the rolling lease gate below.
-      // When added, treat 0/missing the same way: fall through to system default.
+      options = { ...defaults, id: key };
     }
-    const limiter = new Bottleneck({
-      ...defaults,
-      id: key,
-    });
-    // Heartbeat: timestamp every dispatch so the watchdog can tell a healthy
-    // queue (just dispatched a job) from a wedged one (queue has work but
-    // nothing has been dispatched in a while).
-    limiter.on("executing", () => {
-      lastDispatchAt.set(key, Date.now());
+    const limiter = limiterFactory(options);
+    limiterEffectiveSettings.set(limiter, { ...options });
+    limiter.on("queued", () => {
+      limiterWatchdog.noteQueued(key, limiter);
     });
+    const markQueueProgress = () => {
+      limiterWatchdog.noteProgress(key, limiter);
+    };
+    limiter.on("executing", markQueueProgress);
+    // A long-running job can leave older work queued. Start the idle grace
+    // from its completion, not from when that waiting work first arrived.
+    limiter.on("done", markQueueProgress);
 
     limiters.set(key, limiter);
-    lastDispatchAt.set(key, Date.now());
     limiterLastUsed.set(key, Date.now());
   }
 
@@ -618,15 +502,6 @@ function getLimiter(provider, connectionId, model = null) {
   return limiters.get(key);
 }
 
-function evictLimiterAndDropQueued(key: string, limiter: Bottleneck, reason: string): void {
-  if (limiters.get(key) === limiter) {
-    limiters.delete(key);
-    lastDispatchAt.delete(key);
-    limiterLastUsed.delete(key);
-  }
-  trackAsyncOperation(limiter.stop({ dropWaitingJobs: true, dropErrorMessage: reason }));
-}
-
 /**
  * Acquire a rate limit slot before making a request.
  * If rate limiting is disabled for this connection, returns immediately.
@@ -651,20 +526,22 @@ export async function withRateLimit(provider, connectionId, model, fn, signal =
     throw err;
   }
 
-  const maxWaitMs = currentRequestQueueSettings.maxWaitMs;
-  const queueStartedAt = Date.now();
-  const rpmLease = await rpmGate.acquire(
+  // Proactive sliding-window fallback for header-less providers with a declared cap
+  // (Fase 8.2). No-op unless PROVIDER_DEFAULT_RATE_LIMITS has an entry for `provider`.
+  await awaitProviderDefaultSlot(
     provider,
     connectionId,
-    model,
     signal,
-    maxWaitMs,
-    queueStartedAt
+    currentRequestQueueSettings.maxWaitMs
   );
+
   const limiter = getLimiter(provider, connectionId, model);
-  const key = getLimiterKey(provider, connectionId, model);
-  const jobId = `${key}:job-${nextJobTraceId++}`;
-  const scheduleOpts = { id: jobId };
+  // Bottleneck's `expiration` starts only after a job leaves QUEUED. The
+  // legacy maxWaitMs setting therefore bounds limiter-managed execution; it
+  // is not a queue-wait deadline.
+  const executionExpirationMs = currentRequestQueueSettings.maxWaitMs;
+  const scheduleOpts =
+    executionExpirationMs && executionExpirationMs > 0 ? { expiration: executionExpirationMs } : {};
 
   // Issue #6593: opt-in admission cap — fast-reject before Bottleneck's
   // schedule() (and before any downstream compression/prompt work runs) when
@@ -675,129 +552,96 @@ export async function withRateLimit(provider, connectionId, model, fn, signal =
     model ? `${provider}/${model}` : provider
   );
   if (admissionErr) {
-    rpmLease?.release();
     logRateLimit(
       `🚧 [RATE-LIMIT] ${getLimiterKey(provider, connectionId, model)} — queue full, rejecting fast (maxQueueDepth=${currentRequestQueueSettings.maxQueueDepth})`
     );
     throw admissionErr;
   }
 
-  let dispatched = false;
-  let queueExpired = false;
-  let dispatchCancelled = false;
-  let queueTimer: ReturnType<typeof setTimeout> | undefined;
-  const remainingWaitMs =
-    maxWaitMs > 0 ? Math.max(1, maxWaitMs - (Date.now() - queueStartedAt)) : 0;
-  const queueTimeoutPromise =
-    remainingWaitMs > 0
-      ? new Promise<never>((_, reject) => {
-          queueTimer = setTimeout(() => {
-            if (dispatched) return;
-            queueExpired = true;
-            logRateLimit(
-              `⏰ [RATE-LIMIT] ${key} — job exceeded ${Math.ceil(maxWaitMs / 1000)}s queue wait budget, dropping`
-            );
-            reject(new Error("rate-limit-queue-timeout"));
-          }, remainingWaitMs);
-        })
-      : null;
-  const scheduled = limiter.schedule(scheduleOpts, async () => {
-    if (queueExpired) {
-      throw createQueueTimeoutError(provider, model, maxWaitMs);
-    }
-    if (dispatchCancelled) {
-      const error = new Error("The operation was aborted before limiter dispatch");
-      error.name = "AbortError";
-      throw error;
-    }
-    if (signal?.aborted) {
-      const error = new Error("The operation was aborted before limiter dispatch");
-      error.name = "AbortError";
-      throw error;
-    }
-    dispatched = true;
-    if (queueTimer) clearTimeout(queueTimer);
-    return fn();
-  });
-
   try {
     if (signal) {
       let abortListener: (() => void) | undefined;
-      const abortPromise = new Promise<never>((_, reject) => {
-        const onAbort = () => {
-          const reason = signal.reason;
-          // Reject before evicting the queued job so the caller observes its
-          // abort reason instead of Bottleneck's internal drop error.
-          if (reason instanceof Error) {
-            reject(reason);
-          } else {
-            const err = new Error(
-              typeof reason === "string" ? reason : "The operation was aborted"
-            );
-            err.name = "AbortError";
-            if (reason !== undefined) {
-              (err as Error & { cause?: unknown }).cause = reason;
-            }
-            reject(err);
-          }
-          if (!dispatched) {
-            dispatchCancelled = true;
-            if (queueTimer) clearTimeout(queueTimer);
-            // Leave the cancelled job in Bottleneck so queued peers are not dropped.
-            // Its scheduled callback will consume one queue turn and exit before fn().
-          }
-        };
-        if (signal.aborted) {
-          onAbort();
+      const { promise: abortPromise, reject: rejectAbort } = Promise.withResolvers<never>();
+      const onAbort = () => {
+        const reason = signal.reason;
+        // Preserve native Error reasons (including AbortController's
+        // read-only DOMException) instead of mutating or wrapping them.
+        if (reason instanceof Error) {
+          rejectAbort(reason);
           return;
         }
+        const err = new Error(typeof reason === "string" ? reason : "The operation was aborted");
+        err.name = "AbortError";
+        if (reason !== undefined) {
+          (err as Error & { cause?: unknown }).cause = reason;
+        }
+        rejectAbort(err);
+      };
+      if (signal.aborted) {
+        onAbort();
+      } else {
         abortListener = onAbort;
         signal.addEventListener("abort", abortListener, { once: true });
-      });
+      }
 
       try {
-        const races: Promise<unknown>[] = [scheduled, abortPromise];
-        if (queueTimeoutPromise) races.push(queueTimeoutPromise);
-        return await Promise.race(races);
+        return await Promise.race([limiter.schedule(scheduleOpts, fn), abortPromise]);
       } finally {
         if (abortListener) {
           signal.removeEventListener("abort", abortListener);
         }
       }
     } else {
-      return await (queueTimeoutPromise
-        ? Promise.race([scheduled, queueTimeoutPromise])
-        : scheduled);
+      return await limiter.schedule(scheduleOpts, fn);
     }
   } catch (err) {
-    if (queueTimer) clearTimeout(queueTimer);
-    if (!dispatched) rpmLease?.release();
-    if (err?.message === "rate-limit-upstream-429") {
-      const rateLimitErr = new Error(
-        `Request dropped while the ${provider} connection was under an upstream rate-limit cooldown`,
-        { cause: err }
-      ) as Error & { code?: string; status?: number };
-      rateLimitErr.code = "RATE_LIMIT_UPSTREAM_429";
-      rateLimitErr.status = 429;
-      throw rateLimitErr;
+    // Only Bottleneck-owned failures are rewritten. Application code can throw
+    // the same text and must retain its original identity and semantics.
+    if (
+      err instanceof Bottleneck.BottleneckError &&
+      /^This job timed out after \d+ ms\.$/.test(err.message)
+    ) {
+      const key = getLimiterKey(provider, connectionId, model);
+      logRateLimit(
+        `⏰ [RATE-LIMIT] ${key} — limiter-managed execution expired after ${Math.ceil((executionExpirationMs || 0) / 1000)}s`
+      );
+      throw markLocalRateLimitError(
+        new Error(
+          `Request exceeded OmniRoute's local rate-limit execution expiration ` +
+            `(legacy resilienceSettings.requestQueue.maxWaitMs=${executionExpirationMs}ms) for ` +
+            `${model ? `${provider}/${model}` : provider}. Bottleneck applies this deadline only ` +
+            `after dispatch; it does not bound queue wait and is not an upstream-generated timeout.`,
+          { cause: err }
+        ),
+        RATE_LIMIT_EXECUTION_TIMEOUT_CODE
+      );
     }
-    // The watchdog's stop({ dropWaitingJobs: true }) wedge-recovery (above) rejects
-    // queued jobs with this exact message. Rewrite it the same way as the timeout
-    // case — a clear, OmniRoute-owned, classifiable error — so combo's transient-error
-    // handling (which already treats a 502 as retryable) falls back to the next target
-    // immediately instead of surfacing Bottleneck's internal wording.
-    if (err?.message === "rate-limit-watchdog-wedge-reset") {
+
+    if (
+      err instanceof Bottleneck.BottleneckError &&
+      err.message === "rate-limit-watchdog-wedge-reset"
+    ) {
+      const cleanup = limiterWatchdog.getEviction(limiter);
+      if (!cleanup) throw err;
+
+      let cleanupError: unknown;
+      try {
+        await cleanup;
+      } catch (error) {
+        cleanupError = error;
+        errorRateLimit("[RATE-LIMIT] Wedge cleanup failed:", error);
+      }
+
+      const key = getLimiterKey(provider, connectionId, model);
+      logRateLimit(`↪️ [RATE-LIMIT] ${key} — surfacing local wedge; caller will not be replayed`);
       const wedgeErr = new Error(
         `Request dropped: the local rate-limit queue for ${model ? `${provider}/${model}` : provider} ` +
-          `was detected as wedged (stalled with nothing executing) and force-reset. This is OmniRoute's ` +
-          `own queue recovering, not an upstream error.`,
+          `was detected as wedged (stalled with nothing executing) and force-reset. OmniRoute does ` +
+          `not replay dropped work automatically; combo routing may fall back to another target.`,
         { cause: err }
-      ) as Error & { code?: string };
-      wedgeErr.code = "RATE_LIMIT_QUEUE_WEDGED";
-      throw wedgeErr;
-    }
-    if (err?.message === "rate-limit-queue-timeout") {
-      throw createQueueTimeoutError(provider, model, maxWaitMs);
+      ) as Error & { cleanupError?: unknown };
+      if (cleanupError !== undefined) wedgeErr.cleanupError = cleanupError;
+      throw markLocalRateLimitError(wedgeErr, RATE_LIMIT_QUEUE_WEDGED_CODE);
     }
     throw err;
   }
@@ -842,12 +686,21 @@ export function updateFromHeaders(provider, connectionId, headers, status, model
       `🚫 [RATE-LIMIT] ${provider}:${connectionId.slice(0, 8)} — 429 received, pausing for ${Math.ceil(retryAfterMs / 1000)}s, dropping ${counts.QUEUED} queued request(s)`
     );
 
-    rpmGate.block(provider, connectionId, model, retryAfterMs);
-
-    // Evict from the cache before stopping so follow-up requests get a fresh
-    // instance. Stopping the unreachable instance rejects its queued jobs and
-    // releases its heartbeat without poisoning the replacement limiter.
-    evictLimiterAndDropQueued(limiterKey, limiter, "rate-limit-upstream-429");
+    // Evict from the cache so follow-up learning from the same error body
+    // can materialize a fresh limiter immediately. Do NOT call limiter.stop() —
+    // it permanently rejects future .schedule() calls with "This limiter has been stopped".
+    // In-flight requests holding a reference to the evicted instance will fail (they
+    // were already going to fail — the 429 means the API rejected them), but future
+    // requests will get a fresh Bottleneck instance via getLimiter().
+    // Call disconnect() (not stop()) to release Bottleneck's internal heartbeat timer
+    // without permanently poisoning the instance for any remaining in-flight jobs.
+    // Without disconnect() here, every 429 leaks a heartbeat timer until GC reclaims
+    // the abandoned Bottleneck; under sustained quota pressure that is a real leak.
+    limiters.delete(limiterKey);
+    limiterWatchdog.forget(limiter);
+    limiterLastUsed.delete(limiterKey);
+    preservedReplacementSettings.delete(limiterKey);
+    trackAsyncOperation(limiter.disconnect());
     return;
   }
 
@@ -856,41 +709,40 @@ export function updateFromHeaders(provider, connectionId, headers, status, model
     logRateLimit(
       `⚠️ [RATE-LIMIT] ${provider}:${connectionId.slice(0, 8)} — near capacity, slowing down`
     );
-    trackAsyncOperation(applyLimiterSettings(limiter, { minTime: 200 }));
+    updateLimiterSettings(limiter, {
+      minTime: 200, // Add 200ms between requests
+    });
     return;
   }
 
   // Normal response — update limiter from headers
   if (!isNaN(limit) && limit > 0) {
+    const resetMs = parseResetTime(resetStr) || 60000;
+
     // Calculate optimal minTime from RPM limit
     const minTime = Math.max(0, Math.floor(60000 / limit) - 10); // Small buffer
 
     const updates: LimiterUpdateSettings = { minTime };
-    const resetMs = parseResetTime(resetStr) || 60000;
 
-    // Keep adaptive pacing from response headers, but do not mutate an RPM
-    // reservoir. RPM admission is enforced by the rolling lease gate.
+    // If remaining is low (< 10% of limit), set reservoir to throttle immediately
     if (!isNaN(remaining)) {
       if (remaining < limit * 0.1) {
-        rpmGate.learnHeaderWindow(
-          provider,
-          connectionId,
-          model,
-          remaining,
-          resetMs,
-          Date.now() + resetMs
-        );
+        updates.reservoir = remaining;
+        updates.reservoirRefreshAmount = limit;
+        updates.reservoirRefreshInterval = resetMs;
         logRateLimit(
           `⚠️ [RATE-LIMIT] ${provider}:${connectionId.slice(0, 8)} — ${remaining}/${limit} remaining, throttling`
         );
       } else if (remaining > limit * 0.5) {
         // Plenty of headroom — relax the limiter
         updates.minTime = 0;
-        rpmGate.clearLearnedHeaderWindow(provider, connectionId, model);
+        updates.reservoir = null;
+        updates.reservoirRefreshAmount = null;
+        updates.reservoirRefreshInterval = null;
       }
     }
 
-    trackAsyncOperation(applyLimiterSettings(limiter, updates));
+    updateLimiterSettings(limiter, updates);
 
     // Persist learned limits (debounced)
     recordLearnedLimit(
@@ -1003,6 +855,14 @@ export async function __flushLearnedLimitsForTests() {
   }
 }
 
+export function __setLimiterFactoryForTests(factory: LimiterFactory): void {
+  limiterFactory = factory;
+}
+
+export async function __runLimiterWatchdogForTests(now = Date.now()): Promise<void> {
+  await limiterWatchdog.run(now);
+}
+
 export async function __resetRateLimitManagerForTests() {
   if (persistTimer) {
     clearTimeout(persistTimer);
@@ -1019,11 +879,11 @@ export async function __resetRateLimitManagerForTests() {
   }
   limiters.clear();
   enabledConnections.clear();
-  connectionRateLimitOverrides.clear();
-  rpmGate.reset();
   initialized = false;
-  lastDispatchAt.clear();
   limiterLastUsed.clear();
+  preservedReplacementSettings.clear();
+  limiterFactory = defaultLimiterFactory;
+  limiterWatchdog.reset();
   shutdownHandlersRegistered = false;
 
   for (const key of Object.keys(learnedLimits)) {
@@ -1094,7 +954,7 @@ async function loadPersistedLimits() {
         const limiter = limiters.get(key);
         if (limiter && limit > 0) {
           const inferredMinTime = minTime || Math.max(0, Math.floor(60000 / limit) - 10);
-          await applyLimiterSettings(limiter, { minTime: inferredMinTime });
+          updateLimiterSettings(limiter, { minTime: inferredMinTime });
           count++;
         }
       }
@@ -1125,10 +985,15 @@ export function updateFromResponseBody(provider, connectionId, responseBody, sta
   const { retryAfterMs, reason } = parseRetryAfterFromBody(responseBody);
 
   if (retryAfterMs && retryAfterMs > 0) {
-    getLimiter(provider, connectionId, model);
+    const limiter = getLimiter(provider, connectionId, model);
     logRateLimit(
       `🚫 [RATE-LIMIT] ${provider}:${connectionId.slice(0, 8)} — body-parsed retry: ${Math.ceil(retryAfterMs / 1000)}s (${reason})`
     );
-    rpmGate.block(provider, connectionId, model, retryAfterMs);
+
+    updateLimiterSettings(limiter, {
+      reservoir: 0,
+      reservoirRefreshAmount: 60,
+      reservoirRefreshInterval: retryAfterMs,
+    });
   }
 }
diff --git a/open-sse/services/rateLimitManager/admission.ts b/open-sse/services/rateLimitManager/admission.ts
index d9ace7bf13..af3192e0d5 100644
--- a/open-sse/services/rateLimitManager/admission.ts
+++ b/open-sse/services/rateLimitManager/admission.ts
@@ -13,8 +13,10 @@
  * @module services/rateLimitManager/admission
  */
 
+import { markLocalRateLimitError, RATE_LIMIT_QUEUE_FULL_CODE } from "./errors";
+
 export interface QueueFullError extends Error {
-  code: "RATE_LIMIT_QUEUE_FULL";
+  code: typeof RATE_LIMIT_QUEUE_FULL_CODE;
   status: 429;
 }
 
@@ -36,13 +38,8 @@ export function checkQueueAdmission(
       `queued request(s), at or above the configured admission cap maxQueueDepth (${maxQueueDepth}) ` +
       `— this is OmniRoute's request queue (resilienceSettings.requestQueue.maxQueueDepth), not an ` +
       `upstream rejection. Raise it in Settings → Resilience if this is expected burst traffic.`
-  ) as Error & { code?: string; status?: number };
-  err.code = "RATE_LIMIT_QUEUE_FULL";
-  // chatCore's generic catch-all fallback (open-sse/handlers/chatCore.ts) maps a
-  // status-less error to HTTP 502 — which also risks tripping the whole-provider
-  // circuit breaker (PROVIDER_BREAKER_FAILURE_STATUSES includes 502) for what is a
-  // purely local, in-process admission decision. Tag 429 explicitly so it is read
-  // via `error.status` before that fallback kicks in.
-  err.status = 429;
-  return err as QueueFullError;
+  );
+  // The public code/status remain useful to callers, while the WeakMap brand
+  // is the provenance signal used by health and routing decisions.
+  return markLocalRateLimitError(err, RATE_LIMIT_QUEUE_FULL_CODE) as QueueFullError;
 }
diff --git a/open-sse/services/rateLimitManager/errors.ts b/open-sse/services/rateLimitManager/errors.ts
new file mode 100644
index 0000000000..167f92c834
--- /dev/null
+++ b/open-sse/services/rateLimitManager/errors.ts
@@ -0,0 +1,94 @@
+export const RATE_LIMIT_EXECUTION_TIMEOUT_CODE = "RATE_LIMIT_EXECUTION_TIMEOUT";
+export const RATE_LIMIT_QUEUE_FULL_CODE = "RATE_LIMIT_QUEUE_FULL";
+export const RATE_LIMIT_QUEUE_WEDGED_CODE = "RATE_LIMIT_QUEUE_WEDGED";
+export const LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE = "RATE_LIMIT_QUEUE_TIMEOUT";
+
+export type LocalRateLimitErrorCode =
+  | typeof RATE_LIMIT_EXECUTION_TIMEOUT_CODE
+  | typeof RATE_LIMIT_QUEUE_FULL_CODE
+  | typeof RATE_LIMIT_QUEUE_WEDGED_CODE;
+
+export type TrustedLocalRateLimitErrorCode =
+  LocalRateLimitErrorCode | typeof LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE;
+
+export interface TrustedLocalRateLimitFailure {
+  code: TrustedLocalRateLimitErrorCode;
+  status: 429 | 503 | 504;
+}
+
+const localRateLimitErrors = new WeakMap<object, TrustedLocalRateLimitFailure>();
+const localRateLimitResponses = new WeakMap<Response, TrustedLocalRateLimitFailure>();
+
+function getStatusForCode(code: TrustedLocalRateLimitErrorCode): 429 | 503 | 504 {
+  switch (code) {
+    case RATE_LIMIT_QUEUE_FULL_CODE:
+      return 429;
+    case RATE_LIMIT_EXECUTION_TIMEOUT_CODE:
+      return 504;
+    case RATE_LIMIT_QUEUE_WEDGED_CODE:
+    case LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE:
+      return 503;
+  }
+}
+
+/**
+ * Brand an error created by OmniRoute's local limiter. The WeakMap identity,
+ * not the public code string, is the trusted provenance signal.
+ */
+export function markLocalRateLimitError<T extends Error>(
+  error: T,
+  code: TrustedLocalRateLimitErrorCode
+): T & { code: TrustedLocalRateLimitErrorCode; status: 429 | 503 | 504 } {
+  const failure = Object.freeze({ code, status: getStatusForCode(code) });
+  localRateLimitErrors.set(error, failure);
+  const branded = error as T & {
+    code: TrustedLocalRateLimitErrorCode;
+    status: 429 | 503 | 504;
+  };
+  branded.code = failure.code;
+  branded.status = failure.status;
+  return branded;
+}
+
+export function getTrustedLocalRateLimitError(error: unknown): TrustedLocalRateLimitFailure | null {
+  if (!error || (typeof error !== "object" && typeof error !== "function")) return null;
+  return localRateLimitErrors.get(error as object) ?? null;
+}
+
+/**
+ * Return the public fields for a trusted local failure without its low-level
+ * Bottleneck cause, which must remain server-side diagnostic context.
+ */
+export function getClientSafeLocalRateLimitError(
+  error: unknown
+): (TrustedLocalRateLimitFailure & { message: string }) | null {
+  const failure = getTrustedLocalRateLimitError(error);
+  if (!failure) return null;
+  return {
+    ...failure,
+    message: error instanceof Error ? error.message : "Local rate-limit failure",
+  };
+}
+
+/**
+ * Transfer trusted local provenance from a branded error to its generated
+ * internal Response. Provider-controlled bodies and headers cannot set this.
+ */
+export function markTrustedLocalRateLimitResponse(response: Response, error: unknown): Response {
+  const failure = getTrustedLocalRateLimitError(error);
+  if (failure) localRateLimitResponses.set(response, failure);
+  return response;
+}
+
+export function getTrustedLocalRateLimitResponse(
+  response: Response
+): TrustedLocalRateLimitFailure | null {
+  return localRateLimitResponses.get(response) ?? null;
+}
+
+/** Preserve trusted provenance when an internal response wrapper must allocate. */
+export function inheritTrustedLocalRateLimitResponse(source: Response, target: Response): Response {
+  const failure = localRateLimitResponses.get(source);
+  if (failure) localRateLimitResponses.set(target, failure);
+  return target;
+}
diff --git a/open-sse/services/rateLimitManager/wedgeWatchdog.ts b/open-sse/services/rateLimitManager/wedgeWatchdog.ts
new file mode 100644
index 0000000000..624a413880
--- /dev/null
+++ b/open-sse/services/rateLimitManager/wedgeWatchdog.ts
@@ -0,0 +1,210 @@
+import Bottleneck from "bottleneck";
+
+export const WATCHDOG_INTERVAL_MS = 30_000;
+
+const INACTIVE_LIMITER_MS = 10 * 60 * 1000;
+const IDLE_CAPACITY_WEDGE_GRACE_MS = 10_000;
+
+interface IdleCapacitySnapshot {
+  lastProgress: number;
+  reservoir: number | null;
+}
+
+interface LimiterWedgeWatchdogDependencies {
+  limiters: Map<string, Bottleneck>;
+  limiterLastUsed: Map<string, number>;
+  limiterEffectiveSettings: WeakMap<Bottleneck, Bottleneck.ConstructorOptions>;
+  preservedReplacementSettings: Map<string, Bottleneck.ConstructorOptions>;
+  trackBackground: (promise: Promise<unknown>) => void;
+  log: (...args: unknown[]) => void;
+  warn: (...args: unknown[]) => void;
+}
+
+/**
+ * Detects a Bottleneck queue that has remained idle despite immediately usable
+ * capacity. State is keyed by limiter identity so late events from an evicted
+ * instance cannot mutate the replacement's progress record.
+ */
+export class LimiterWedgeWatchdog {
+  private queueProgressAt = new WeakMap<Bottleneck, number>();
+  private evictions = new WeakMap<Bottleneck, Promise<boolean>>();
+  private currentRun: Promise<void> | null = null;
+
+  constructor(private readonly dependencies: LimiterWedgeWatchdogDependencies) {}
+
+  noteQueued(key: string, limiter: Bottleneck): void {
+    if (this.dependencies.limiters.get(key) !== limiter) return;
+    if (!this.queueProgressAt.has(limiter)) this.queueProgressAt.set(limiter, Date.now());
+  }
+
+  noteProgress(key: string, limiter: Bottleneck): void {
+    if (this.dependencies.limiters.get(key) !== limiter) return;
+    if (limiter.counts().QUEUED > 0) {
+      this.queueProgressAt.set(limiter, Date.now());
+    } else {
+      this.queueProgressAt.delete(limiter);
+    }
+  }
+
+  forget(limiter: Bottleneck): void {
+    this.queueProgressAt.delete(limiter);
+  }
+
+  getEviction(limiter: Bottleneck): Promise<boolean> | undefined {
+    return this.evictions.get(limiter);
+  }
+
+  run(now = Date.now()): Promise<void> {
+    if (this.currentRun) return this.currentRun;
+    const run = this.tick(now);
+    this.currentRun = run;
+    void run.then(
+      () => {
+        if (this.currentRun === run) this.currentRun = null;
+      },
+      () => {
+        if (this.currentRun === run) this.currentRun = null;
+      }
+    );
+    return run;
+  }
+
+  reset(): void {
+    this.queueProgressAt = new WeakMap();
+    this.evictions = new WeakMap();
+    this.currentRun = null;
+  }
+
+  private async tick(now: number): Promise<void> {
+    const { limiters, limiterLastUsed, log, trackBackground, warn } = this.dependencies;
+
+    for (const [key, limiter] of Array.from(limiters)) {
+      const lastUsed = limiterLastUsed.get(key) ?? 0;
+      if (now - lastUsed <= INACTIVE_LIMITER_MS) continue;
+
+      const counts = limiter.counts();
+      if (counts.QUEUED > 0 || counts.RUNNING > 0 || counts.EXECUTING > 0) continue;
+
+      limiters.delete(key);
+      this.queueProgressAt.delete(limiter);
+      limiterLastUsed.delete(key);
+      log(
+        `[RATE-LIMIT] Evicting idle limiter: ${key} ` +
+          `(inactive for ${Math.round((now - lastUsed) / 1000)}s)`
+      );
+      trackBackground(limiter.disconnect());
+    }
+
+    for (const [key, limiter] of Array.from(limiters)) {
+      const snapshot = await this.getStableIdleCapacity(key, limiter, now);
+      if (!snapshot) continue;
+
+      const counts = limiter.counts();
+      const cleanup = this.evict(key, limiter, snapshot);
+      if (!cleanup) continue;
+
+      warn(
+        `[RATE-LIMIT] WEDGED: ${key} queued=${counts.QUEUED} running=0 executing=0 ` +
+          `stalled=${now - snapshot.lastProgress}ms with idle capacity — force-resetting`
+      );
+      await cleanup;
+    }
+  }
+
+  private async getStableIdleCapacity(
+    key: string,
+    limiter: Bottleneck,
+    now: number
+  ): Promise<IdleCapacitySnapshot | null> {
+    const before = limiter.counts();
+    if (before.QUEUED === 0) {
+      this.queueProgressAt.delete(limiter);
+      return null;
+    }
+    if (before.RUNNING > 0 || before.EXECUTING > 0) return null;
+
+    const lastProgress = this.queueProgressAt.get(limiter);
+    if (lastProgress === undefined) {
+      this.queueProgressAt.set(limiter, now);
+      return null;
+    }
+    if (now - lastProgress < IDLE_CAPACITY_WEDGE_GRACE_MS) return null;
+
+    let canRunNow: boolean;
+    let reservoir: number | null;
+    try {
+      // Every job this manager submits has Bottleneck's default weight of 1.
+      // check(1) is an eligibility query for exactly that shape, not a generic
+      // query about an arbitrary weighted queue head.
+      canRunNow = await limiter.check(1);
+      if (!canRunNow) return null;
+      reservoir = await limiter.currentReservoir();
+    } catch {
+      return null;
+    }
+    if (this.dependencies.limiters.get(key) !== limiter) return null;
+
+    const after = limiter.counts();
+    if (
+      after.QUEUED === 0 ||
+      after.RUNNING > 0 ||
+      after.EXECUTING > 0 ||
+      this.queueProgressAt.get(limiter) !== lastProgress
+    ) {
+      return null;
+    }
+    return { lastProgress, reservoir };
+  }
+
+  private evict(
+    key: string,
+    limiter: Bottleneck,
+    snapshot: IdleCapacitySnapshot
+  ): Promise<boolean> | null {
+    const { limiterEffectiveSettings, limiterLastUsed, limiters, preservedReplacementSettings } =
+      this.dependencies;
+    if (limiters.get(key) !== limiter) return null;
+
+    const counts = limiter.counts();
+    if (
+      counts.QUEUED === 0 ||
+      counts.RUNNING > 0 ||
+      counts.EXECUTING > 0 ||
+      this.queueProgressAt.get(limiter) !== snapshot.lastProgress
+    ) {
+      return null;
+    }
+
+    const effectiveSettings = limiterEffectiveSettings.get(limiter) ?? {};
+    preservedReplacementSettings.set(key, {
+      ...effectiveSettings,
+      id: key,
+      // Carry consumed capacity forward. Restarting the refresh interval from
+      // replacement creation is conservative and cannot grant an early burst.
+      reservoir: snapshot.reservoir,
+    });
+    limiters.delete(key);
+    this.queueProgressAt.delete(limiter);
+    limiterLastUsed.delete(key);
+
+    // Register this Promise before stop() runs. Every dropped caller awaits the
+    // same cleanup and is surfaced exactly once; none is replayed automatically.
+    const stopped = Promise.resolve().then(() =>
+      limiter.stop({
+        dropWaitingJobs: true,
+        dropErrorMessage: "rate-limit-watchdog-wedge-reset",
+      })
+    );
+    const cleanup = stopped
+      .then(
+        () => limiter.disconnect(),
+        async (stopError: unknown) => {
+          await limiter.disconnect();
+          throw stopError;
+        }
+      )
+      .then(() => true);
+    this.evictions.set(limiter, cleanup);
+    return cleanup;
+  }
+}
diff --git a/src/lib/api/modelTestRunner.ts b/src/lib/api/modelTestRunner.ts
index e3041e7c1f..bb2aeb8dba 100644
--- a/src/lib/api/modelTestRunner.ts
+++ b/src/lib/api/modelTestRunner.ts
@@ -17,6 +17,7 @@ import {
   isDailyQuotaExhausted,
 } from "@omniroute/open-sse/services/accountFallback";
 import { looksLikeQuotaExhausted } from "@/shared/utils/classify429";
+import { getTrustedLocalRateLimitError } from "@omniroute/open-sse/services/rateLimitManager/errors";
 
 const INTERNAL_ORIGIN = "http://omniroute.internal";
 export const DEFAULT_MODEL_TEST_TIMEOUT_MS = 30_000;
@@ -484,12 +485,14 @@ export async function runSingleModelTest(
         rateLimited: true,
       };
     }
+    const localRateLimitFailure = getTrustedLocalRateLimitError(error);
     return {
       modelId: fullModelStr,
-      status: "error",
+      status: localRateLimitFailure?.status === 429 ? "rate_limited" : "error",
       latencyMs,
-      httpStatus: 500,
+      httpStatus: localRateLimitFailure?.status ?? 500,
       error: getErrorMessage(error),
+      ...(localRateLimitFailure?.status === 429 ? { rateLimited: true } : {}),
     };
   }
   let latencyMs = Date.now() - startTime;
diff --git a/src/lib/resilience/settings/types.ts b/src/lib/resilience/settings/types.ts
index 8e2ca28723..0bc16d48b2 100644
--- a/src/lib/resilience/settings/types.ts
+++ b/src/lib/resilience/settings/types.ts
@@ -16,6 +16,10 @@ export interface RequestQueueSettings {
   requestsPerMinute: number;
   minTimeBetweenRequestsMs: number;
   concurrentRequests: number;
+  /**
+   * Legacy persisted key used as Bottleneck's post-dispatch execution
+   * expiration. It does not bound time spent in Bottleneck's QUEUED state.
+   */
   maxWaitMs: number;
   /**
    * Issue #6593: opt-in admission cap on the local rate-limit queue. When the
diff --git a/src/sse/handlers/chatHelpers.ts b/src/sse/handlers/chatHelpers.ts
index 835c0d58e6..ddfdaee039 100644
--- a/src/sse/handlers/chatHelpers.ts
+++ b/src/sse/handlers/chatHelpers.ts
@@ -17,6 +17,7 @@ import {
   providerCircuitOpenResponse,
   unavailableResponse,
 } from "@omniroute/open-sse/utils/error.ts";
+import { inheritTrustedLocalRateLimitResponse } from "@omniroute/open-sse/services/rateLimitManager/errors.ts";
 import { HTTP_STATUS } from "@omniroute/open-sse/config/constants.ts";
 import {
   runWithProxyContext,
@@ -898,7 +899,7 @@ export function withSessionHeader(response: Response, sessionId: string | null):
       headers: response.headers,
     });
     cloned.headers.set("X-OmniRoute-Session-Id", sessionId);
-    return cloned;
+    return inheritTrustedLocalRateLimitResponse(response, cloned);
   }
 }
 
@@ -915,7 +916,7 @@ export function withCorrelationId(response: Response, correlationId: string | nu
       headers: response.headers,
     });
     cloned.headers.set("X-Correlation-Id", correlationId);
-    return cloned;
+    return inheritTrustedLocalRateLimitResponse(response, cloned);
   }
 }
 
@@ -960,6 +961,6 @@ export function withSelectedConnectionHeader(
       headers: response.headers,
     });
     cloned.headers.set("X-OmniRoute-Selected-Connection-Id", connectionId);
-    return cloned;
+    return inheritTrustedLocalRateLimitResponse(response, cloned);
   }
 }
diff --git a/src/sse/handlers/chatPredicates.ts b/src/sse/handlers/chatPredicates.ts
index 2f8c8e5bd4..cbff12d506 100644
--- a/src/sse/handlers/chatPredicates.ts
+++ b/src/sse/handlers/chatPredicates.ts
@@ -1,5 +1,6 @@
 import { isLocalStreamLifecycleError } from "../../shared/utils/circuitBreaker";
 import { isRequestScopedUpstreamFailure } from "./comboFailureLogging";
+import { getTrustedLocalRateLimitResponse } from "@omniroute/open-sse/services/rateLimitManager/errors";
 
 export const PROVIDER_BREAKER_FAILURE_STATUSES = new Set([408, 500, 502, 503, 504]);
 
@@ -12,7 +13,13 @@ export function isProviderBreakerFailureStatus(status: number): boolean {
 // otherwise a client abort (502 default, error='request_signal_aborted') trips the
 // provider-wide breaker. Pure predicate, unit-testable without the full request path.
 export function shouldTripProviderBreakerForResult(
-  result: { status: number; errorCode?: string | null; errorType?: string | null; error?: unknown },
+  result: {
+    status: number;
+    response?: Response;
+    errorCode?: string | null;
+    errorType?: string | null;
+    error?: unknown;
+  },
   isCombo: boolean,
   forceLiveComboTest: boolean
 ): boolean {
@@ -20,6 +27,7 @@ export function shouldTripProviderBreakerForResult(
     !forceLiveComboTest &&
     !isCombo &&
     !isRequestScopedUpstreamFailure({ code: result.errorCode, type: result.errorType }) &&
+    !(result.response && getTrustedLocalRateLimitResponse(result.response)) &&
     !isLocalStreamLifecycleError(result.error) &&
     PROVIDER_BREAKER_FAILURE_STATUSES.has(Number(result.status))
   );
diff --git a/stryker.conf.json b/stryker.conf.json
index 7a0cc3dfcc..b512eedb28 100644
--- a/stryker.conf.json
+++ b/stryker.conf.json
@@ -280,6 +280,8 @@
       "tests/unit/quota-streaming-consumption-usd.test.ts",
       "tests/unit/qwen-web-content-array-serialization.test.ts",
       "tests/unit/rate-limit-enhanced.test.ts",
+      "tests/unit/rate-limit-execution-timeout-message-4165.test.ts",
+      "tests/unit/rate-limit-local-error-classification.test.ts",
       "tests/unit/rate-limit-manager.test.ts",
       "tests/unit/rate-limit-queue-timeout-lockout.test.ts",
       "tests/unit/repro-7503-no-choices.test.ts",
diff --git a/tests/unit/model-test-runner.test.ts b/tests/unit/model-test-runner.test.ts
index 4ca0bc9472..bca51db207 100644
--- a/tests/unit/model-test-runner.test.ts
+++ b/tests/unit/model-test-runner.test.ts
@@ -10,6 +10,13 @@ import {
   resolveModelTestTimeoutMs,
   classifyTestErrorQuota,
 } from "@/lib/api/modelTestRunner.ts";
+import Bottleneck from "bottleneck";
+import * as rateLimitManager from "@omniroute/open-sse/services/rateLimitManager.ts";
+import {
+  markLocalRateLimitError,
+  RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+  RATE_LIMIT_QUEUE_WEDGED_CODE,
+} from "@omniroute/open-sse/services/rateLimitManager/errors.ts";
 
 // ---------------------------------------------------------------------------
 // parseRetryAfterHeader — Retry-After is either delta-seconds or an HTTP-date.
@@ -375,3 +382,46 @@ test("classifyTestErrorQuota: daily-quota wins over credits-exhausted (isTransie
   assert.equal(result.isQuota, true);
   assert.equal(result.isTransient, true);
 });
+test("runSingleModelTest preserves trusted local limiter HTTP statuses", async () => {
+  const connection = await createProviderConnection({
+    provider: "openai",
+    authType: "apikey",
+    name: "model-test-local-limiter-errors",
+    apiKey: "sk-model-test-local-limiter-errors",
+    isActive: true,
+    testStatus: "active",
+  });
+
+  try {
+    for (const [code, status] of [
+      [RATE_LIMIT_QUEUE_WEDGED_CODE, 503],
+      [RATE_LIMIT_EXECUTION_TIMEOUT_CODE, 504],
+    ] as const) {
+      await rateLimitManager.__resetRateLimitManagerForTests();
+      rateLimitManager.enableRateLimitProtection(connection.id);
+      rateLimitManager.__setLimiterFactoryForTests((options) => {
+        const limiter = new Bottleneck(options);
+        Object.defineProperty(limiter, "schedule", {
+          configurable: true,
+          value: async () => {
+            throw markLocalRateLimitError(new Error(`trusted ${code}`), code);
+          },
+        });
+        return limiter;
+      });
+
+      const result = await runSingleModelTest({
+        providerId: "openai",
+        modelId: "gpt-4o",
+        connectionId: connection.id,
+        timeoutMs: 5_000,
+      });
+
+      assert.equal(result.status, "error");
+      assert.equal(result.httpStatus, status);
+      assert.equal(result.error, `Error: trusted ${code}`);
+    }
+  } finally {
+    await rateLimitManager.__resetRateLimitManagerForTests();
+  }
+});
diff --git a/tests/unit/rate-limit-execution-timeout-message-4165.test.ts b/tests/unit/rate-limit-execution-timeout-message-4165.test.ts
new file mode 100644
index 0000000000..3cae26ebbd
--- /dev/null
+++ b/tests/unit/rate-limit-execution-timeout-message-4165.test.ts
@@ -0,0 +1,129 @@
+/**
+ * #4165 — classify Bottleneck's execution expiration accurately.
+ *
+ * OmniRoute passes the legacy `requestQueue.maxWaitMs` value to Bottleneck as
+ * the job `expiration`. Bottleneck starts that timer only after a job leaves
+ * QUEUED, so it bounds limiter-managed execution and does not bound queue wait.
+ *
+ * The raw Bottleneck message (`This job timed out after <N> ms.`) still needs an
+ * OmniRoute-owned code and message so it cannot masquerade as an upstream-
+ * generated timeout. The original error remains available as `.cause`.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-rl-execution-timeout-"));
+process.env.DATA_DIR = TEST_DATA_DIR;
+
+// Dynamic imports are required because DATA_DIR must be set before DB modules evaluate.
+const core = await import("../../src/lib/db/core.ts");
+const resilienceSettings = await import("../../src/lib/resilience/settings.ts");
+const rateLimitManager = await import("../../open-sse/services/rateLimitManager.ts");
+const { getClientSafeLocalRateLimitError, getTrustedLocalRateLimitError } =
+  await import("../../open-sse/services/rateLimitManager/errors.ts");
+const { formatProviderError } = await import("../../open-sse/utils/error.ts");
+
+// This contract test deliberately drives Bottleneck's real expiration timer.
+function wait(ms: number) {
+  const { promise, resolve } = Promise.withResolvers<void>();
+  setTimeout(resolve, ms);
+  return promise;
+}
+
+test.afterEach(async () => {
+  await rateLimitManager.__resetRateLimitManagerForTests();
+});
+
+test.after(() => {
+  core.resetDbInstance();
+  fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+});
+
+// Drive a real Bottleneck execution expiration with a function that outlives it.
+async function triggerExecutionExpiration() {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    concurrentRequests: 1,
+    requestsPerMinute: 100000,
+    minTimeBetweenRequestsMs: 0,
+    maxWaitMs: 40,
+  });
+  rateLimitManager.enableRateLimitProtection("conn-execution-timeout");
+
+  return rateLimitManager.withRateLimit("openai", "conn-execution-timeout", "gpt-4o", async () => {
+    await wait(400); // > maxWaitMs (40ms) → Bottleneck fails the job
+    return "should-not-reach";
+  });
+}
+
+test("#4165 execution expiration is local and accurately named", async () => {
+  let caught: (Error & { code?: string; cause?: { message?: string } }) | undefined;
+  try {
+    await triggerExecutionExpiration();
+    assert.fail("expected the limiter-managed execution to expire");
+  } catch (err) {
+    caught = err as Error & { code?: string; cause?: { message?: string } };
+  }
+  assert.ok(caught, "an error should have been thrown");
+
+  assert.equal(
+    caught.code,
+    "RATE_LIMIT_EXECUTION_TIMEOUT",
+    "error must carry the local execution-expiration code"
+  );
+
+  assert.match(caught.message, /execution expiration/i);
+  assert.match(caught.message, /does not bound queue wait/i);
+  assert.match(
+    caught.message,
+    /not an upstream-generated timeout/i,
+    "message should explicitly disclaim an upstream-generated timeout"
+  );
+  assert.doesNotMatch(
+    caught.message,
+    /This job timed out/,
+    "raw Bottleneck/upstream-looking string must not leak into the surfaced message"
+  );
+
+  // The original Bottleneck error is preserved for debugging.
+  assert.ok(caught.cause, "original error should be preserved as cause");
+  assert.match(String(caught.cause?.message ?? ""), /This job timed out/);
+
+  assert.deepEqual(getTrustedLocalRateLimitError(caught), {
+    code: "RATE_LIMIT_EXECUTION_TIMEOUT",
+    status: 504,
+  });
+  const safeError = getClientSafeLocalRateLimitError(caught);
+  assert.ok(safeError);
+  const clientMessage = formatProviderError(safeError, "openai", "gpt-4o", 504);
+  assert.match(clientMessage, /execution expiration/i);
+  assert.doesNotMatch(
+    clientMessage,
+    /This job timed out/,
+    "client and call-log formatting must not append the retained Bottleneck cause"
+  );
+});
+
+test("#4165 a job that completes within the execution expiration is unaffected", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    concurrentRequests: 1,
+    requestsPerMinute: 100000,
+    minTimeBetweenRequestsMs: 0,
+    maxWaitMs: 5000,
+  });
+  rateLimitManager.enableRateLimitProtection("conn-fast");
+
+  const result = await rateLimitManager.withRateLimit(
+    "openai",
+    "conn-fast",
+    "gpt-4o",
+    async () => "ok"
+  );
+  assert.equal(result, "ok");
+});
diff --git a/tests/unit/rate-limit-local-error-classification.test.ts b/tests/unit/rate-limit-local-error-classification.test.ts
new file mode 100644
index 0000000000..aad8ebb57d
--- /dev/null
+++ b/tests/unit/rate-limit-local-error-classification.test.ts
@@ -0,0 +1,355 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-rl-local-errors-"));
+process.env.DATA_DIR = TEST_DATA_DIR;
+process.env.API_KEY_SECRET = "test-rate-limit-local-error-secret";
+
+// Dynamic imports are required because DATA_DIR must be set before DB modules evaluate.
+const core = await import("../../src/lib/db/core.ts");
+const providersDb = await import("../../src/lib/db/providers.ts");
+const { handleComboChat } = await import("../../open-sse/services/combo.ts");
+const {
+  isComboRequestScopedFailure,
+  isRequestScopedUpstreamFailure,
+  shouldRecordProviderBreakerFailure,
+  shouldSkipConnDisable,
+} = await import("../../open-sse/services/combo/comboPredicates.ts");
+const {
+  LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE,
+  RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+  RATE_LIMIT_QUEUE_WEDGED_CODE,
+  getTrustedLocalRateLimitError,
+  getTrustedLocalRateLimitResponse,
+  inheritTrustedLocalRateLimitResponse,
+  markLocalRateLimitError,
+  markTrustedLocalRateLimitResponse,
+} = await import("../../open-sse/services/rateLimitManager/errors.ts");
+const accountFallback = await import("../../open-sse/services/accountFallback.ts");
+const providerCooldown = await import("../../open-sse/services/providerCooldownTracker.ts");
+const rateLimitSemaphore = await import("../../open-sse/services/rateLimitSemaphore.ts");
+const { createStreamingErrorResult } =
+  await import("../../open-sse/handlers/chatCore/streamErrorResult.ts");
+const { shouldTripProviderBreakerForResult } =
+  await import("../../src/sse/handlers/chatPredicates.ts");
+
+const LOCAL_ERROR_MESSAGE = "OmniRoute repaired a local limiter queue";
+
+function createLocalLimiterSseResponse(connectionId: string, code = RATE_LIMIT_QUEUE_WEDGED_CODE) {
+  const error = markLocalRateLimitError(new Error(LOCAL_ERROR_MESSAGE), code);
+  const { response } = createStreamingErrorResult(
+    getTrustedLocalRateLimitError(error)?.status ?? 503,
+    LOCAL_ERROR_MESSAGE,
+    code,
+    "rate_limit_queue_wedged"
+  );
+  response.headers.set("X-OmniRoute-Selected-Connection-Id", connectionId);
+  return markTrustedLocalRateLimitResponse(response, error);
+}
+
+function createUpstreamCollisionResponse(connectionId: string) {
+  return new Response(
+    JSON.stringify({
+      error: {
+        message: "Provider emitted a colliding code",
+        code: RATE_LIMIT_QUEUE_WEDGED_CODE,
+        type: "rate_limit_queue_wedged",
+      },
+    }),
+    {
+      status: 503,
+      headers: {
+        "content-type": "application/json",
+        "X-OmniRoute-Selected-Connection-Id": connectionId,
+      },
+    }
+  );
+}
+
+function createSuccessResponse(connectionId: string) {
+  return new Response(JSON.stringify({ choices: [{ message: { content: "fallback ok" } }] }), {
+    status: 200,
+    headers: {
+      "content-type": "application/json",
+      "X-OmniRoute-Selected-Connection-Id": connectionId,
+    },
+  });
+}
+
+const log = { info() {}, warn() {}, error() {}, debug() {} };
+const settings = {
+  modelLockout: {
+    enabled: true,
+    errorCodes: [503],
+    baseCooldownMs: 3_000,
+    maxCooldownMs: 5_000,
+    maxBackoffSteps: 10,
+    useExponentialBackoff: true,
+  },
+};
+
+test.afterEach(() => {
+  accountFallback.clearAllModelLockouts();
+  accountFallback.clearProviderFailure("openai");
+  providerCooldown.clearCooldownState();
+  rateLimitSemaphore.resetAll();
+  core.resetDbInstance();
+  fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+  fs.mkdirSync(TEST_DATA_DIR, { recursive: true });
+});
+
+test.after(() => {
+  accountFallback.clearAllModelLockouts();
+  accountFallback.clearProviderFailure("openai");
+  providerCooldown.clearCooldownState();
+  rateLimitSemaphore.resetAll();
+  core.resetDbInstance();
+  fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+});
+
+test("trusted provenance, not a public code string, classifies local limiter failures", () => {
+  const executionError = markLocalRateLimitError(
+    new Error("local execution expiration"),
+    RATE_LIMIT_EXECUTION_TIMEOUT_CODE
+  );
+  const localResponse = markTrustedLocalRateLimitResponse(
+    new Response("local", { status: 504 }),
+    executionError
+  );
+  const collisionResponse = createUpstreamCollisionResponse("collision-conn");
+
+  assert.deepEqual(getTrustedLocalRateLimitError(executionError), {
+    code: RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+    status: 504,
+  });
+  assert.equal(getTrustedLocalRateLimitResponse(localResponse)?.status, 504);
+  const wrappedResponse = inheritTrustedLocalRateLimitResponse(
+    localResponse,
+    new Response("wrapped local", { status: 504 })
+  );
+  assert.equal(getTrustedLocalRateLimitResponse(wrappedResponse)?.status, 504);
+  assert.equal(
+    isRequestScopedUpstreamFailure({ code: RATE_LIMIT_EXECUTION_TIMEOUT_CODE }),
+    false,
+    "an upstream-controlled code string must not establish local provenance"
+  );
+  assert.equal(
+    isComboRequestScopedFailure(localResponse, "local execution expiration", {
+      code: RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+    }),
+    true
+  );
+  assert.equal(
+    isComboRequestScopedFailure(collisionResponse, "provider collision", {
+      code: RATE_LIMIT_QUEUE_WEDGED_CODE,
+    }),
+    false
+  );
+  assert.equal(
+    shouldTripProviderBreakerForResult(
+      {
+        status: 504,
+        response: localResponse,
+        errorCode: RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+      },
+      false,
+      false
+    ),
+    false
+  );
+  assert.equal(
+    shouldTripProviderBreakerForResult(
+      {
+        status: 503,
+        response: collisionResponse,
+        errorCode: RATE_LIMIT_QUEUE_WEDGED_CODE,
+      },
+      false,
+      false
+    ),
+    true,
+    "an untrusted upstream collision must remain a provider-health failure"
+  );
+  assert.equal(
+    shouldSkipConnDisable(
+      {
+        status: 504,
+        response: localResponse,
+        errorCode: RATE_LIMIT_EXECUTION_TIMEOUT_CODE,
+      },
+      false,
+      false,
+      "openai"
+    ),
+    true
+  );
+  assert.equal(
+    shouldSkipConnDisable(
+      {
+        status: 503,
+        response: collisionResponse,
+        errorCode: RATE_LIMIT_QUEUE_WEDGED_CODE,
+      },
+      false,
+      false,
+      "openai"
+    ),
+    false
+  );
+  assert.equal(
+    shouldRecordProviderBreakerFailure({
+      isStreamReadinessFailure: false,
+      status: 504,
+      sameProviderNext: false,
+      skipProviderBreaker: false,
+      requestScopedFailure: true,
+      error: executionError,
+      isProxyUnreachable: false,
+    }),
+    false
+  );
+});
+
+test("legacy queue-timeout compatibility also requires trusted local provenance", () => {
+  const untrusted = new Response("legacy collision", { status: 503 });
+  const legacyError = markLocalRateLimitError(
+    new Error("legacy local timeout"),
+    LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE
+  );
+  const trusted = markTrustedLocalRateLimitResponse(
+    new Response("legacy local timeout", { status: 503 }),
+    legacyError
+  );
+
+  assert.equal(
+    isComboRequestScopedFailure(untrusted, "legacy collision", {
+      code: LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE,
+    }),
+    false
+  );
+  assert.equal(
+    isComboRequestScopedFailure(trusted, "legacy local timeout", {
+      code: LEGACY_RATE_LIMIT_QUEUE_TIMEOUT_CODE,
+    }),
+    true
+  );
+});
+
+for (const strategy of ["priority", "round-robin"] as const) {
+  test(`${strategy} fallback preserves all health state for a trusted local SSE failure`, async () => {
+    const connection = await providersDb.createProviderConnection({
+      provider: "openai",
+      authType: "apikey",
+      name: `local-wedge-${strategy}`,
+      apiKey: `sk-local-wedge-${strategy}`,
+      isActive: true,
+      testStatus: "active",
+      rateLimitedUntil: null,
+      backoffLevel: 0,
+      providerSpecificData: {},
+    });
+    const models = [
+      {
+        kind: "model",
+        model: "openai/gpt-local-first",
+        connectionId: connection.id,
+      },
+      {
+        kind: "model",
+        model: "openai/gpt-local-second",
+        connectionId: connection.id,
+      },
+    ];
+    const calls: string[] = [];
+
+    const result = await handleComboChat({
+      body: {},
+      combo: {
+        name: `local-wedge-${strategy}-combo`,
+        strategy,
+        models,
+        config: {
+          maxRetries: 1,
+          retryDelayMs: 0,
+          fallbackDelayMs: 0,
+          maxConcurrency: 1,
+        },
+      },
+      handleSingleModel: async (_body, modelStr) => {
+        calls.push(modelStr);
+        return calls.length === 1
+          ? createLocalLimiterSseResponse(connection.id)
+          : createSuccessResponse(connection.id);
+      },
+      isModelAvailable: async () => true,
+      log,
+      settings,
+      allCombos: null,
+    });
+
+    assert.equal(result.status, 200, `attempted targets: ${calls.join(", ")}`);
+    assert.deepEqual(calls, ["openai/gpt-local-first", "openai/gpt-local-second"]);
+    assert.equal(accountFallback.isModelLocked("openai", connection.id, "gpt-local-first"), false);
+    assert.equal(
+      accountFallback.getProviderBreakerState("openai")?.failureCount ?? 0,
+      0,
+      "local failure must not increment the provider breaker"
+    );
+    assert.equal(
+      providerCooldown.isProviderInCooldown("openai", connection.id),
+      false,
+      "local failure must not enter provider cooldown"
+    );
+    const semaphoreStates = Object.values(rateLimitSemaphore.getStats());
+    assert.equal(
+      semaphoreStates.some((state) => state.rateLimitedUntil !== null),
+      false,
+      "local failure must not cool a round-robin semaphore"
+    );
+    const storedConnection = await providersDb.getProviderConnectionById(connection.id);
+    assert.equal(storedConnection?.testStatus, "active");
+    assert.equal(storedConnection?.rateLimitedUntil ?? null, null);
+  });
+}
+
+test("an untrusted upstream code collision retains ordinary health penalties", async () => {
+  const connection = await providersDb.createProviderConnection({
+    provider: "openai",
+    authType: "apikey",
+    name: "upstream-local-code-collision",
+    apiKey: "sk-upstream-local-code-collision",
+    isActive: true,
+    testStatus: "active",
+    providerSpecificData: {},
+  });
+
+  const result = await handleComboChat({
+    body: {},
+    combo: {
+      name: "upstream-local-code-collision-combo",
+      strategy: "priority",
+      models: [
+        {
+          kind: "model",
+          model: "openai/gpt-collision",
+          connectionId: connection.id,
+        },
+      ],
+      config: { maxRetries: 1, retryDelayMs: 0, fallbackDelayMs: 0 },
+    },
+    handleSingleModel: async () => createUpstreamCollisionResponse(connection.id),
+    isModelAvailable: async () => true,
+    log,
+    settings,
+    allCombos: null,
+  });
+
+  assert.equal(result.status, 503);
+  assert.ok(
+    (accountFallback.getProviderBreakerState("openai")?.failureCount ?? 0) >= 1,
+    "the upstream 503 must remain eligible for provider-breaker accounting"
+  );
+});
diff --git a/tests/unit/rate-limit-manager.test.ts b/tests/unit/rate-limit-manager.test.ts
index 1d2ec645e1..b1d411e62a 100644
--- a/tests/unit/rate-limit-manager.test.ts
+++ b/tests/unit/rate-limit-manager.test.ts
@@ -1,6 +1,5 @@
 import test from "node:test";
 import assert from "node:assert/strict";
-import Bottleneck from "bottleneck";
 import fs from "node:fs";
 import os from "node:os";
 import path from "node:path";
@@ -8,19 +7,86 @@ import path from "node:path";
 const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-rate-limit-manager-"));
 process.env.DATA_DIR = TEST_DATA_DIR;
 
+// Dynamic imports are required because DATA_DIR must be set before DB modules evaluate.
 const core = await import("../../src/lib/db/core.ts");
 const providersDb = await import("../../src/lib/db/providers.ts");
 const resilienceSettings = await import("../../src/lib/resilience/settings.ts");
 const rateLimitManager = await import("../../open-sse/services/rateLimitManager.ts");
+const rateLimitErrors = await import("../../open-sse/services/rateLimitManager/errors.ts");
 const accountFallback = await import("../../open-sse/services/accountFallback.ts");
+const Bottleneck = (await import("bottleneck")).default;
 
+// These integration-style tests exercise real Bottleneck timer/event behavior.
 function wait(ms) {
-  return new Promise((resolve) => setTimeout(resolve, ms));
+  const { promise, resolve } = Promise.withResolvers<void>();
+  setTimeout(resolve, ms);
+  return promise;
+}
+
+// A real deadline is intentional: these tests drive real Bottleneck queues, and
+// a broken cleanup path otherwise leaves Node's test process pending forever.
+async function settleWithin<T>(
+  promise: Promise<T>,
+  message: string,
+  timeoutMs = 2_000
+): Promise<T> {
+  let timeout: NodeJS.Timeout;
+  const deadline = new Promise<never>((_, reject) => {
+    timeout = setTimeout(() => reject(new Error(message)), timeoutMs);
+  });
+  try {
+    return await Promise.race([promise, deadline]);
+  } finally {
+    clearTimeout(timeout);
+  }
+}
+
+type TestBottleneck = InstanceType<typeof Bottleneck> & {
+  _drainAll: (...args: unknown[]) => Promise<unknown>;
+};
+
+/**
+ * Fault injection for the observed Bottleneck failure mode: jobs enter the real
+ * Bottleneck queue, but its internal drain loop stops making progress. Keep the
+ * private mutation in this one helper so the tests otherwise exercise public
+ * manager and Bottleneck behavior.
+ */
+function injectDrainWedge(limiter: InstanceType<typeof Bottleneck>): TestBottleneck {
+  const wedged = limiter as TestBottleneck;
+  wedged._drainAll = () => Promise.resolve(null);
+  return wedged;
+}
+
+async function waitForCondition(
+  condition: () => boolean | Promise<boolean>,
+  message: string
+): Promise<void> {
+  const deadline = Date.now() + 1_000;
+  while (!(await condition())) {
+    if (Date.now() >= deadline) throw new Error(message);
+    await wait(5);
+  }
+}
+
+async function expectWedgeError(promise: Promise<unknown>): Promise<void> {
+  await assert.rejects(
+    settleWithin(promise, "stranded limiter caller did not reject after wedge recovery"),
+    (error: Error & { code?: string }) => {
+      assert.equal(error.code, "RATE_LIMIT_QUEUE_WEDGED");
+      assert.deepEqual(rateLimitErrors.getTrustedLocalRateLimitError(error), {
+        code: "RATE_LIMIT_QUEUE_WEDGED",
+        status: 503,
+      });
+      return true;
+    }
+  );
 }
 
 async function flushBackgroundWork() {
   await wait(50);
-  await new Promise((resolve) => setImmediate(resolve));
+  const { promise, resolve } = Promise.withResolvers<void>();
+  setImmediate(resolve);
+  await promise;
 }
 
 async function resetStorage() {
@@ -60,296 +126,561 @@ test("rate limit manager bypasses disabled connections and exposes inactive stat
   assert.deepEqual(rateLimitManager.getAllRateLimitStatus(), {});
 });
 
-test("queue expiry does not invoke the provider after a late dispatch", async () => {
+test("idle-capacity watchdog honors grace, cleans up in order, and rejects the stranded caller", async () => {
   await rateLimitManager.applyRequestQueueSettings({
     ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
     autoEnableApiKeyProviders: false,
-    maxWaitMs: 100,
+    maxWaitMs: 240_000,
     requestsPerMinute: 0,
     concurrentRequests: 1,
     minTimeBetweenRequestsMs: 0,
     maxQueueDepth: 0,
   });
 
-  rateLimitManager.enableRateLimitProtection("queue-expiry-conn");
-  let resolveFirstStarted: () => void = () => undefined;
-  const firstStarted = new Promise<void>((resolve) => {
-    resolveFirstStarted = resolve;
+  const cleanupEvents: string[] = [];
+  let limitersCreated = 0;
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    const limiter = new Bottleneck(options);
+    limitersCreated++;
+    if (limitersCreated === 1) {
+      injectDrainWedge(limiter);
+      const originalStop = limiter.stop.bind(limiter);
+      const originalDisconnect = limiter.disconnect.bind(limiter);
+      limiter.stop = async (stopOptions) => {
+        cleanupEvents.push("stop:start");
+        await originalStop(stopOptions);
+        cleanupEvents.push("stop:done");
+      };
+      limiter.disconnect = async (flush) => {
+        cleanupEvents.push("disconnect");
+        await originalDisconnect(flush);
+      };
+    }
+    return limiter;
   });
-  const first = rateLimitManager.withRateLimit(
+
+  rateLimitManager.enableRateLimitProtection("idle-capacity-conn");
+  let executions = 0;
+  const pending = rateLimitManager.withRateLimit(
     "openai",
-    "queue-expiry-conn",
+    "idle-capacity-conn",
     "gpt-4o",
     async () => {
-      resolveFirstStarted();
-      await wait(300);
-      return "first";
+      executions++;
+      return "must-not-run";
     }
   );
-  await firstStarted;
 
-  let secondCalls = 0;
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "queue-expiry-conn", "gpt-4o", async () => {
-      secondCalls++;
-      return "late";
-    }),
-    (error: { code?: string }) => error.code === "RATE_LIMIT_QUEUE_TIMEOUT"
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", "idle-capacity-conn").queued === 1,
+    "the injected drain failure never established a real queued job"
+  );
+  const queuedObservedAt = Date.now();
+
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(queuedObservedAt + 9_000),
+    "watchdog grace-period scan did not finish"
+  );
+  assert.equal(
+    rateLimitManager.getRateLimitStatus("openai", "idle-capacity-conn").queued,
+    1,
+    "the queue must survive before the 10s stability grace"
   );
 
-  await first;
-  await wait(50);
-  assert.equal(secondCalls, 0, "a queue-expired job must not invoke the provider later");
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(queuedObservedAt + 11_000),
+    "watchdog wedge cleanup did not finish"
+  );
+  await expectWedgeError(pending);
+
+  assert.equal(executions, 0, "watchdog recovery must never replay application work");
+  assert.equal(limitersCreated, 1, "dropped callers must not create a replacement limiter");
+  assert.deepEqual(cleanupEvents, ["stop:start", "stop:done", "disconnect"]);
 });
 
-test("queue expiry does not drop other queued jobs", async () => {
+test("wedge eviction rejects every queued caller and preserves learned state for future traffic", async () => {
   await rateLimitManager.applyRequestQueueSettings({
     ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
     autoEnableApiKeyProviders: false,
-    maxWaitMs: 500,
-    requestsPerMinute: 0,
-    concurrentRequests: 1,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 60,
+    concurrentRequests: 6,
     minTimeBetweenRequestsMs: 0,
     maxQueueDepth: 0,
   });
 
-  rateLimitManager.enableRateLimitProtection("queue-peer-conn");
-  let resolveFirstExecuting: () => void = () => undefined;
-  const firstExecuting = new Promise<void>((resolve) => {
-    resolveFirstExecuting = resolve;
-  });
-  let releaseFirst: () => void = () => undefined;
-  const first = rateLimitManager.withRateLimit("openai", "queue-peer-conn", null, async () => {
-    resolveFirstExecuting();
-    await new Promise<void>((resolve) => {
-      releaseFirst = resolve;
-    });
-    return "first";
-  });
-  await firstExecuting;
-
-  const second = rateLimitManager.withRateLimit(
-    "openai",
-    "queue-peer-conn",
-    null,
-    async () => "expired"
-  );
-  await wait(400);
-
-  let thirdCalls = 0;
-  const third = rateLimitManager.withRateLimit("openai", "queue-peer-conn", null, async () => {
-    thirdCalls++;
-    return "third";
-  });
-  await assert.rejects(
-    second,
-    (error: { code?: string }) => error.code === "RATE_LIMIT_QUEUE_TIMEOUT"
-  );
-  releaseFirst();
-  await Promise.all([first, third]);
-  assert.equal(thirdCalls, 1, "a peer queued job must survive another job's expiry");
-});
-
-test("global RPM lease is shared across enabled provider connections", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    maxWaitMs: 1000,
-    requestsPerMinute: 2,
-    concurrentRequests: 10,
-    minTimeBetweenRequestsMs: 0,
-    maxQueueDepth: 0,
+  const createdOptions: Bottleneck.ConstructorOptions[] = [];
+  const createdLimiters: InstanceType<typeof Bottleneck>[] = [];
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    createdOptions.push({ ...options });
+    const limiter = new Bottleneck(options);
+    createdLimiters.push(limiter);
+    if (createdLimiters.length === 1) injectDrainWedge(limiter);
+    return limiter;
   });
 
-  rateLimitManager.enableRateLimitProtection("global-rpm-a");
-  rateLimitManager.enableRateLimitProtection("global-rpm-b");
-  let calls = 0;
-  await rateLimitManager.withRateLimit("openai", "global-rpm-a", null, async () => {
-    calls++;
-  });
-  await rateLimitManager.withRateLimit("anthropic", "global-rpm-b", null, async () => {
-    calls++;
-  });
-
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "global-rpm-a", null, async () => {
-      calls++;
-    }),
-    (error: { code?: string }) => error.code === "RATE_LIMIT_QUEUE_TIMEOUT"
-  );
-  assert.equal(calls, 2, "the global lease blocks the third dispatch across providers");
-});
-
-test("provider/account RPM lease failure does not consume the global lease", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    maxWaitMs: 1000,
-    requestsPerMinute: 2,
-    concurrentRequests: 10,
-    minTimeBetweenRequestsMs: 0,
-    maxQueueDepth: 0,
-  });
-
-  rateLimitManager.enableRateLimitProtection("provider-rpm-a");
-  rateLimitManager.enableRateLimitProtection("provider-rpm-b");
-  rateLimitManager.refreshConnectionRateLimits("provider-rpm-a", { rpm: 1 });
-
-  let calls = 0;
-  await rateLimitManager.withRateLimit("openai", "provider-rpm-a", null, async () => {
-    calls++;
-  });
-
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "provider-rpm-a", null, async () => {
-      calls++;
-    }),
-    (error: { code?: string }) => error.code === "RATE_LIMIT_QUEUE_TIMEOUT"
-  );
-
-  await rateLimitManager.withRateLimit("anthropic", "provider-rpm-b", null, async () => {
-    calls++;
-  });
-  assert.equal(calls, 2, "the failed provider lease did not consume the second global lease");
-});
-
-test("aborted queued work releases its pre-dispatch RPM lease", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    maxWaitMs: 1000,
-    requestsPerMinute: 2,
-    concurrentRequests: 1,
-    minTimeBetweenRequestsMs: 0,
-    maxQueueDepth: 0,
-  });
-
-  rateLimitManager.enableRateLimitProtection("abort-lease-conn");
-  rateLimitManager.enableRateLimitProtection("abort-lease-other");
-  let resolveFirstExecuting: () => void = () => undefined;
-  const firstExecuting = new Promise<void>((resolve) => {
-    resolveFirstExecuting = resolve;
-  });
-  let releaseFirst: () => void = () => undefined;
-  const firstStarted = new Promise<void>((resolve) => {
-    releaseFirst = resolve;
-  });
-  const first = rateLimitManager.withRateLimit("openai", "abort-lease-conn", null, async () => {
-    resolveFirstExecuting();
-    await firstStarted;
-    return "first";
-  });
-  await firstExecuting;
-
-  const controller = new AbortController();
-  let abortedCalls = 0;
-  const queued = rateLimitManager.withRateLimit(
-    "openai",
-    "abort-lease-conn",
-    null,
-    async () => {
-      abortedCalls++;
-      return "should-not-dispatch";
-    },
-    controller.signal
-  );
-  await wait(20);
-  controller.abort();
-  await assert.rejects(queued, (error: { name?: string }) => error.name === "AbortError");
-
-  let thirdCalls = 0;
-  await rateLimitManager.withRateLimit("anthropic", "abort-lease-other", null, async () => {
-    thirdCalls++;
-  });
-  releaseFirst();
-  await first;
-  assert.equal(abortedCalls, 0, "aborted queued work must not invoke the provider");
-  assert.equal(thirdCalls, 1, "aborted work must return its unused global lease");
-});
-
-test("aborting one queued request does not drop queued peers", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    maxWaitMs: 1000,
-    requestsPerMinute: 0,
-    concurrentRequests: 1,
-    minTimeBetweenRequestsMs: 0,
-    maxQueueDepth: 0,
-  });
-
-  const connectionId = "abort-peer-conn";
+  const connectionId = "learned-state-conn";
   rateLimitManager.enableRateLimitProtection(connectionId);
-  let resolveFirstExecuting: () => void = () => undefined;
-  const firstExecuting = new Promise<void>((resolve) => {
-    resolveFirstExecuting = resolve;
-  });
-  let releaseFirst: () => void = () => undefined;
-  const first = rateLimitManager.withRateLimit("test-provider", connectionId, null, async () => {
-    resolveFirstExecuting();
-    await new Promise<void>((resolve) => {
-      releaseFirst = resolve;
-    });
-  });
-
-  await firstExecuting;
-  const limiter = rateLimitManager.__getLimiterForTests("test-provider", connectionId);
-  const controller = new AbortController();
-  let abortedCalls = 0;
-  const aborted = rateLimitManager.withRateLimit(
-    "test-provider",
+  rateLimitManager.updateFromHeaders(
+    "openai",
     connectionId,
-    null,
-    async () => {
-      abortedCalls++;
+    {
+      "x-ratelimit-limit-requests": "100",
+      "x-ratelimit-remaining-requests": "1",
+      "x-ratelimit-reset-requests": "60s",
     },
-    controller.signal
+    200,
+    "gpt-4o"
+  );
+  await waitForCondition(
+    async () =>
+      (await rateLimitManager.__getLimiterStateForTests("openai", connectionId, "gpt-4o"))
+        ?.reservoir === 1,
+    "the learned reservoir was not applied"
   );
-  let peerCalls = 0;
-  const peer = rateLimitManager.withRateLimit("test-provider", connectionId, null, async () => {
-    peerCalls++;
-  });
-  for (let attempt = 0; attempt < 200 && limiter.counts().QUEUED < 2; attempt++) {
-    await wait(5);
-  }
-  assert.ok(limiter.counts().QUEUED >= 2, "both queued requests must be present before abort");
-  controller.abort();
-  await assert.rejects(aborted, (error: { name?: string }) => error.name === "AbortError");
 
-  releaseFirst();
-  await Promise.all([first, peer]);
-  assert.equal(abortedCalls, 0, "aborted queued work must not invoke the provider");
-  assert.equal(peerCalls, 1);
+  let executions = 0;
+  const stranded = Array.from({ length: 3 }, () =>
+    rateLimitManager.withRateLimit("openai", connectionId, "gpt-4o", async () => {
+      executions++;
+      return "must-not-run";
+    })
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 3,
+    "all callers did not enter the wedged queue"
+  );
+
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 11_000),
+    "multi-caller wedge cleanup did not finish"
+  );
+  const settled = await settleWithin(
+    Promise.allSettled(stranded),
+    "not every stranded limiter caller settled"
+  );
+  assert.equal(executions, 0);
+  assert.equal(createdLimiters.length, 1, "wedge recovery must not retry any dropped caller");
+  for (const result of settled) {
+    assert.equal(result.status, "rejected");
+    assert.equal((result as PromiseRejectedResult).reason.code, "RATE_LIMIT_QUEUE_WEDGED");
+  }
+
+  assert.equal(
+    await rateLimitManager.withRateLimit("openai", connectionId, "gpt-4o", async () => "future"),
+    "future"
+  );
+  assert.equal(createdLimiters.length, 2, "future traffic should create one replacement limiter");
+  assert.equal(createdOptions[1].reservoir, 1, "replacement must retain the remaining reservoir");
+  assert.equal(createdOptions[1].minTime, 590, "replacement must retain learned request spacing");
+
+  const queuedAfterPreservedPermit = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "after-refill"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 1,
+    "the preserved reservoir should allow only one request"
+  );
+  await createdLimiters[1].incrementReservoir(1);
+  assert.equal(await queuedAfterPreservedPermit, "after-refill");
 });
 
-test("dispatched provider failures retain their RPM lease", async () => {
+test("global settings changed after eviction replace stale pending configuration", async () => {
+  const connection = await providersDb.createProviderConnection({
+    provider: "openai",
+    authType: "apikey",
+    name: "wedge-global-settings",
+    apiKey: "sk-wedge-global-settings",
+    isActive: true,
+    rateLimitProtection: true,
+  });
   await rateLimitManager.applyRequestQueueSettings({
     ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
     autoEnableApiKeyProviders: false,
-    maxWaitMs: 500,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 60,
+    concurrentRequests: 6,
+    minTimeBetweenRequestsMs: 0,
+  });
+
+  const createdOptions: Bottleneck.ConstructorOptions[] = [];
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    createdOptions.push({ ...options });
+    const limiter = new Bottleneck(options);
+    if (createdOptions.length === 1) injectDrainWedge(limiter);
+    return limiter;
+  });
+
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    connection.id,
+    "gpt-4o",
+    async () => "must-not-run"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connection.id).queued === 1,
+    "global-settings caller did not enter the wedged queue"
+  );
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 11_000),
+    "global-settings wedge cleanup did not finish"
+  );
+  await expectWedgeError(pending);
+
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 8,
+    concurrentRequests: 3,
+    minTimeBetweenRequestsMs: 31,
+  });
+  assert.equal(
+    await rateLimitManager.withRateLimit(
+      "openai",
+      connection.id,
+      "gpt-4o",
+      async () => "new-policy"
+    ),
+    "new-policy"
+  );
+  assert.equal(createdOptions[1].reservoir, 8);
+  assert.equal(createdOptions[1].maxConcurrent, 3);
+  assert.equal(createdOptions[1].minTime, 31);
+});
+
+test("connection overrides changed after eviction replace stale pending configuration", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 60,
+    concurrentRequests: 6,
+    minTimeBetweenRequestsMs: 0,
+  });
+  const createdOptions: Bottleneck.ConstructorOptions[] = [];
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    createdOptions.push({ ...options });
+    const limiter = new Bottleneck(options);
+    if (createdOptions.length === 1) injectDrainWedge(limiter);
+    return limiter;
+  });
+
+  const connectionId = "wedge-override-conn";
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "must-not-run"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 1,
+    "override caller did not enter the wedged queue"
+  );
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 11_000),
+    "override wedge cleanup did not finish"
+  );
+  await expectWedgeError(pending);
+
+  rateLimitManager.refreshConnectionRateLimits(connectionId, {
+    rpm: 7,
+    maxConcurrent: 2,
+    minTime: 25,
+  });
+  assert.equal(
+    await rateLimitManager.withRateLimit(
+      "openai",
+      connectionId,
+      "gpt-4o",
+      async () => "new-override"
+    ),
+    "new-override"
+  );
+  assert.equal(createdOptions[1].reservoir, 7);
+  assert.equal(createdOptions[1].maxConcurrent, 2);
+  assert.equal(createdOptions[1].minTime, 25);
+});
+
+test("disable and re-enable discard learned state preserved by an earlier wedge", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 60,
+    concurrentRequests: 6,
+    minTimeBetweenRequestsMs: 0,
+  });
+  const createdOptions: Bottleneck.ConstructorOptions[] = [];
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    createdOptions.push({ ...options });
+    const limiter = new Bottleneck(options);
+    if (createdOptions.length === 1) injectDrainWedge(limiter);
+    return limiter;
+  });
+
+  const connectionId = "wedge-reenabled-conn";
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  rateLimitManager.updateFromHeaders(
+    "openai",
+    connectionId,
+    {
+      "x-ratelimit-limit-requests": "100",
+      "x-ratelimit-remaining-requests": "1",
+      "x-ratelimit-reset-requests": "60s",
+    },
+    200,
+    "gpt-4o"
+  );
+  await waitForCondition(
+    async () =>
+      (await rateLimitManager.__getLimiterStateForTests("openai", connectionId, "gpt-4o"))
+        ?.reservoir === 1,
+    "learned reservoir was not applied before disable/re-enable"
+  );
+
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "must-not-run"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 1,
+    "disable/re-enable caller did not enter the wedged queue"
+  );
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 11_000),
+    "disable/re-enable wedge cleanup did not finish"
+  );
+  await expectWedgeError(pending);
+
+  rateLimitManager.disableRateLimitProtection(connectionId);
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  assert.equal(
+    await rateLimitManager.withRateLimit("openai", connectionId, "gpt-4o", async () => "reenabled"),
+    "reenabled"
+  );
+  assert.equal(createdOptions[1].reservoir, 60);
+  assert.equal(createdOptions[1].minTime, 0);
+});
+
+test("idle-capacity watchdog preserves a legitimate exhausted-reservoir queue", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
     requestsPerMinute: 1,
-    concurrentRequests: 10,
+    concurrentRequests: 1,
     minTimeBetweenRequestsMs: 0,
     maxQueueDepth: 0,
   });
 
-  rateLimitManager.enableRateLimitProtection("failed-dispatch-a");
-  rateLimitManager.enableRateLimitProtection("failed-dispatch-b");
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "failed-dispatch-a", null, async () => {
-      throw new Error("upstream failure");
-    }),
-    /upstream failure/
+  let limiter: TestBottleneck | null = null;
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    limiter = new Bottleneck(options) as TestBottleneck;
+    return limiter;
+  });
+  rateLimitManager.enableRateLimitProtection("zero-reservoir-conn");
+  assert.equal(
+    await rateLimitManager.withRateLimit(
+      "openai",
+      "zero-reservoir-conn",
+      "gpt-4o",
+      async () => "first"
+    ),
+    "first"
   );
 
-  let secondCalls = 0;
-  await assert.rejects(
-    rateLimitManager.withRateLimit("anthropic", "failed-dispatch-b", null, async () => {
-      secondCalls++;
-    }),
-    (error: { code?: string }) => error.code === "RATE_LIMIT_QUEUE_TIMEOUT"
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    "zero-reservoir-conn",
+    "gpt-4o",
+    async () => "after-refresh"
   );
-  assert.equal(secondCalls, 0, "a dispatched failure still counts against the RPM window");
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", "zero-reservoir-conn").queued === 1,
+    "the exhausted reservoir did not queue the follow-up"
+  );
+
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 150_000),
+    "zero-reservoir watchdog scan did not finish"
+  );
+  assert.equal(
+    rateLimitManager.getRateLimitStatus("openai", "zero-reservoir-conn").queued,
+    1,
+    "a zero-reservoir wait must survive regardless of elapsed time"
+  );
+
+  assert.ok(limiter);
+  await limiter.incrementReservoir(1);
+  assert.equal(await pending, "after-refresh");
+});
+
+test("idle-capacity watchdog preserves a real Bottleneck minTime wait", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 0,
+    concurrentRequests: 1,
+    minTimeBetweenRequestsMs: 100,
+    maxQueueDepth: 0,
+  });
+
+  rateLimitManager.enableRateLimitProtection("min-time-conn");
+  await rateLimitManager.withRateLimit("openai", "min-time-conn", "gpt-4o", async () => "first");
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    "min-time-conn",
+    "gpt-4o",
+    async () => "after-min-time"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", "min-time-conn").running === 1,
+    "Bottleneck did not place the minTime-delayed job in RUNNING"
+  );
+
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 150_000),
+    "minTime watchdog scan did not finish"
+  );
+  assert.equal(
+    rateLimitManager.getRateLimitStatus("openai", "min-time-conn").running,
+    1,
+    "a legitimate RUNNING minTime delay must not be evicted"
+  );
+  assert.equal(await pending, "after-min-time");
+});
+
+test("events from an evicted limiter cannot erase replacement queue progress", async () => {
+  await rateLimitManager.applyRequestQueueSettings({
+    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
+    autoEnableApiKeyProviders: false,
+    maxWaitMs: 240_000,
+    requestsPerMinute: 0,
+    concurrentRequests: 1,
+    minTimeBetweenRequestsMs: 0,
+    maxQueueDepth: 0,
+  });
+
+  const limiters: InstanceType<typeof Bottleneck>[] = [];
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    const limiter = new Bottleneck(options);
+    limiters.push(limiter);
+    if (limiters.length === 2) injectDrainWedge(limiter);
+    return limiter;
+  });
+
+  const connectionId = "stale-listener-conn";
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  const { promise: oldGate, resolve: releaseOld } = Promise.withResolvers<void>();
+  const oldExecuting = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => {
+      await oldGate;
+      return "old-first";
+    }
+  );
+  await waitForCondition(
+    () => limiters[0]?.counts().EXECUTING === 1,
+    "the old limiter did not begin executing"
+  );
+  const oldQueued = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "old-second"
+  );
+  await waitForCondition(
+    () => limiters[0]?.counts().QUEUED === 1,
+    "the old limiter did not queue its second job"
+  );
+
+  rateLimitManager.refreshConnectionRateLimits(connectionId, {});
+  const replacementPending = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "must-not-run"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 1,
+    "the replacement limiter did not establish its queue"
+  );
+
+  releaseOld();
+  assert.equal(await oldExecuting, "old-first");
+  assert.equal(await oldQueued, "old-second");
+
+  await settleWithin(
+    rateLimitManager.__runLimiterWatchdogForTests(Date.now() + 11_000),
+    "stale-listener watchdog cleanup did not finish"
+  );
+  await expectWedgeError(replacementPending);
+});
+
+test("watchdog ticks are serialized while an eligibility check is in flight", async () => {
+  const { promise: checkGate, resolve: releaseCheck } = Promise.withResolvers<void>();
+  let checks = 0;
+  rateLimitManager.__setLimiterFactoryForTests((options) => {
+    const limiter = injectDrainWedge(new Bottleneck(options));
+    const originalCheck = limiter.check.bind(limiter);
+    limiter.check = async (weight) => {
+      checks++;
+      await checkGate;
+      return originalCheck(weight);
+    };
+    return limiter;
+  });
+
+  const connectionId = "serialized-watchdog-conn";
+  rateLimitManager.enableRateLimitProtection(connectionId);
+  const pending = rateLimitManager.withRateLimit(
+    "openai",
+    connectionId,
+    "gpt-4o",
+    async () => "must-not-run"
+  );
+  await waitForCondition(
+    () => rateLimitManager.getRateLimitStatus("openai", connectionId).queued === 1,
+    "the serialized-watchdog fixture did not queue"
+  );
+
+  const now = Date.now() + 11_000;
+  const firstTick = rateLimitManager.__runLimiterWatchdogForTests(now);
+  const secondTick = rateLimitManager.__runLimiterWatchdogForTests(now);
+  await waitForCondition(() => checks === 1, "the first tick did not reach limiter.check()");
+  releaseCheck();
+  await settleWithin(
+    Promise.all([firstTick, secondTick]),
+    "serialized watchdog scans did not finish"
+  );
+  await expectWedgeError(pending);
+  assert.equal(checks, 1, "overlapping watchdog calls must share one scan");
+});
+
+test("application errors resembling Bottleneck failures remain untouched", async () => {
+  rateLimitManager.enableRateLimitProtection("lookalike-error-conn");
+  for (const message of [
+    "This job timed out after 240000 ms.",
+    "rate-limit-watchdog-wedge-reset",
+  ]) {
+    const applicationError = new Error(message);
+    await assert.rejects(
+      rateLimitManager.withRateLimit("openai", "lookalike-error-conn", "gpt-4o", async () => {
+        throw applicationError;
+      }),
+      (error) => error === applicationError
+    );
+  }
 });
 
 test("withRateLimit forwards AbortController DOMException without mutating it", async () => {
@@ -462,130 +793,6 @@ test("rate limit manager handles 429 limiter teardown and disable cleanup", asyn
   assert.equal(rateLimitManager.getRateLimitStatus("gemini", "conn-disable").active, false);
 });
 
-test("rate limit manager blocks admission after an upstream 429 retry hint", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    concurrentRequests: 1,
-    requestsPerMinute: 0,
-    maxWaitMs: 100,
-  });
-  rateLimitManager.enableRateLimitProtection("conn-429-block");
-  rateLimitManager.updateFromHeaders(
-    "openai",
-    "conn-429-block",
-    { "retry-after": "1s" },
-    429,
-    "gpt-4o"
-  );
-
-  let providerCalls = 0;
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "conn-429-block", "gpt-4o", async () => {
-      providerCalls++;
-    }),
-    (error: unknown) => {
-      const code = error && typeof error === "object" && "code" in error ? error.code : undefined;
-      assert.equal(code, "RATE_LIMIT_QUEUE_TIMEOUT");
-      assert.match(String((error as Error).message), /upstream rate-limit cooldown/);
-      return true;
-    }
-  );
-  assert.equal(providerCalls, 0);
-});
-
-test("rate limit manager blocks a zero-remaining header window until reset", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    concurrentRequests: 1,
-    requestsPerMinute: 0,
-    maxWaitMs: 100,
-  });
-  rateLimitManager.enableRateLimitProtection("conn-zero-remaining");
-  rateLimitManager.updateFromHeaders(
-    "openai",
-    "conn-zero-remaining",
-    {
-      "x-ratelimit-limit-requests": "10",
-      "x-ratelimit-remaining-requests": "0",
-      "x-ratelimit-reset-requests": "1s",
-    },
-    200
-  );
-
-  let providerCalls = 0;
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "conn-zero-remaining", null, async () => {
-      providerCalls++;
-    }),
-    (error: unknown) => {
-      const code = error && typeof error === "object" && "code" in error ? error.code : undefined;
-      assert.equal(code, "RATE_LIMIT_QUEUE_TIMEOUT");
-      assert.match(String((error as Error).message), /upstream rate-limit cooldown/);
-      return true;
-    }
-  );
-  assert.equal(providerCalls, 0);
-});
-
-test("rate limit manager keeps learned header windows model-scoped where limiters are model-scoped", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    concurrentRequests: 1,
-    requestsPerMinute: 0,
-    maxWaitMs: 100,
-  });
-  rateLimitManager.enableRateLimitProtection("conn-model-header");
-  rateLimitManager.updateFromHeaders(
-    "github",
-    "conn-model-header",
-    {
-      "x-ratelimit-limit-requests": "10",
-      "x-ratelimit-remaining-requests": "0",
-      "x-ratelimit-reset-requests": "1s",
-    },
-    200,
-    "model-a"
-  );
-
-  let providerCalls = 0;
-  await rateLimitManager.withRateLimit("github", "conn-model-header", "model-b", async () => {
-    providerCalls++;
-  });
-  assert.equal(providerCalls, 1);
-});
-
-test("rate limit watchdog resets a queued limiter with received work", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    concurrentRequests: 1,
-    requestsPerMinute: 0,
-    maxWaitMs: 5_000,
-  });
-  rateLimitManager.enableRateLimitProtection("conn-wedge");
-  const limiter = new Bottleneck({ reservoir: 0, id: "test-provider:conn-wedge" });
-  rateLimitManager.__installLimiterForTests("test-provider", "conn-wedge", limiter);
-
-  let providerCalls = 0;
-  const pending = rateLimitManager.withRateLimit("test-provider", "conn-wedge", null, async () => {
-    providerCalls++;
-  });
-  await wait(100);
-  const counts = limiter.counts();
-  assert.equal(counts.RECEIVED, 0);
-  assert.ok(counts.QUEUED > 0);
-
-  rateLimitManager.__setLastDispatchAtForTests(
-    "test-provider",
-    "conn-wedge",
-    null,
-    Date.now() - 120_001
-  );
-  rateLimitManager.__runRateLimitWatchdogForTests();
-
-  await assert.rejects(pending, (error: unknown) => {
-    const code = error && typeof error === "object" && "code" in error ? error.code : undefined;
-    assert.equal(code, "RATE_LIMIT_QUEUE_WEDGED");
-    return true;
-  });
-  assert.equal(providerCalls, 0);
-});
-
 test("rate limit manager uses model-scoped limiter keys for GitHub Copilot (#1624)", async () => {
   rateLimitManager.enableRateLimitProtection("conn-github");
   rateLimitManager.updateFromHeaders(
@@ -617,11 +824,6 @@ test("rate limit manager uses model-scoped limiter keys for GitHub Copilot (#162
 });
 
 test("rate limit manager parses retry hints from response bodies and locks models", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    maxWaitMs: 100,
-    requestsPerMinute: 0,
-  });
   rateLimitManager.enableRateLimitProtection("conn-body");
   rateLimitManager.updateFromResponseBody(
     "openai",
@@ -645,20 +847,6 @@ test("rate limit manager parses retry hints from response bodies and locks model
   assert.equal(limiterState?.key, "openai:conn-body");
   assert.equal(rateLimitManager.getRateLimitStatus("openai", "conn-body").active, true);
 
-  let providerCalls = 0;
-  await assert.rejects(
-    rateLimitManager.withRateLimit("openai", "conn-body", "gpt-4o", async () => {
-      providerCalls++;
-    }),
-    (error: unknown) => {
-      const code = error && typeof error === "object" && "code" in error ? error.code : undefined;
-      assert.equal(code, "RATE_LIMIT_QUEUE_TIMEOUT");
-      assert.match(String((error as Error).message), /upstream rate-limit cooldown/);
-      return true;
-    }
-  );
-  assert.equal(providerCalls, 0);
-
   rateLimitManager.updateFromResponseBody(
     "openai",
     "conn-body",
@@ -762,21 +950,18 @@ test("withRateLimit rejects cleanly when the caller aborts with the default DOME
     isActive: true,
   });
   rateLimitManager.enableRateLimitProtection(String(connection.id));
-
   const controller = new AbortController();
-  // Mirror how a real executor call behaves: it settles once the signal it
-  // was handed aborts, so this job doesn't dangle forever in Bottleneck once
-  // withRateLimit's own Promise.race settles via the abort path below.
-  const settlesOnAbort = (signal) =>
-    new Promise((_resolve, reject) => {
-      signal.addEventListener("abort", () => reject(signal.reason), { once: true });
-    });
-
   const pending = rateLimitManager.withRateLimit(
     "openai",
     String(connection.id),
     "gpt-4o",
-    () => settlesOnAbort(controller.signal),
+    () => {
+      const { promise, reject } = Promise.withResolvers<never>();
+      controller.signal.addEventListener("abort", () => reject(controller.signal.reason), {
+        once: true,
+      });
+      return promise;
+    },
     controller.signal
   );
 
diff --git a/tests/unit/rate-limit-queue-timeout-message-4165.test.ts b/tests/unit/rate-limit-queue-timeout-message-4165.test.ts
deleted file mode 100644
index 685c01e67d..0000000000
--- a/tests/unit/rate-limit-queue-timeout-message-4165.test.ts
+++ /dev/null
@@ -1,148 +0,0 @@
-/**
- * #4165 — surface a clear error when the request-queue (Bottleneck) drops a job.
- *
- * Queue waiting is bounded by a separate timer. Bottleneck's job expiration is
- * intentionally not used because it measures the entire scheduled lifetime and
- * would kill an already-dispatched provider call that is making progress.
- *
- * The queue-only timer still rewrites pre-dispatch expiry into a clear,
- * OmniRoute-owned error that names the knob (`resilienceSettings.requestQueue.maxWaitMs`)
- * and explicitly says it is NOT an upstream timeout.
- */
-import test from "node:test";
-import assert from "node:assert/strict";
-import fs from "node:fs";
-import os from "node:os";
-import path from "node:path";
-
-const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-rl-queue-timeout-"));
-process.env.DATA_DIR = TEST_DATA_DIR;
-
-const core = await import("../../src/lib/db/core.ts");
-const resilienceSettings = await import("../../src/lib/resilience/settings.ts");
-const rateLimitManager = await import("../../open-sse/services/rateLimitManager.ts");
-
-function wait(ms: number) {
-  return new Promise((resolve) => setTimeout(resolve, ms));
-}
-
-// Leave enough scheduling headroom for a loaded CI/devbox while keeping the
-// executing callback longer than the queue-only budget. The actual queued-job
-// case stays short because it controls dispatch deterministically.
-const DISPATCHED_QUEUE_BUDGET_MS = 2_000;
-const QUEUED_QUEUE_BUDGET_MS = 250;
-
-test.afterEach(async () => {
-  await rateLimitManager.__resetRateLimitManagerForTests();
-});
-
-test.after(() => {
-  core.resetDbInstance();
-  fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
-});
-
-// A dispatched provider call may run longer than maxWaitMs without being killed.
-async function triggerQueueTimeout() {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    concurrentRequests: 1,
-    requestsPerMinute: 100000,
-    minTimeBetweenRequestsMs: 0,
-    maxWaitMs: DISPATCHED_QUEUE_BUDGET_MS,
-  });
-  const connectionId = "conn-dispatched-timeout";
-  rateLimitManager.enableRateLimitProtection(connectionId);
-
-  let dispatched = false;
-  const result = await rateLimitManager.withRateLimit(
-    "test-provider",
-    connectionId,
-    null,
-    async () => {
-      dispatched = true;
-      await wait(DISPATCHED_QUEUE_BUDGET_MS + 250);
-      return "should-not-reach";
-    }
-  );
-  return { dispatched, result };
-}
-
-async function triggerQueuedTimeout() {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    concurrentRequests: 1,
-    requestsPerMinute: 0,
-    minTimeBetweenRequestsMs: 0,
-    maxWaitMs: QUEUED_QUEUE_BUDGET_MS,
-  });
-  const connectionId = "conn-queued-timeout";
-  rateLimitManager.enableRateLimitProtection(connectionId);
-
-  let resolveFirstExecuting: () => void = () => undefined;
-  const firstExecuting = new Promise<void>((resolve) => {
-    resolveFirstExecuting = resolve;
-  });
-  let releaseFirst: () => void = () => undefined;
-  const first = rateLimitManager.withRateLimit("test-provider", connectionId, null, async () => {
-    resolveFirstExecuting();
-    await new Promise<void>((resolve) => {
-      releaseFirst = resolve;
-    });
-  });
-  await firstExecuting;
-
-  let caught: unknown;
-  let queuedDispatched = false;
-  try {
-    await rateLimitManager.withRateLimit("test-provider", connectionId, null, async () => {
-      queuedDispatched = true;
-      return "should-not-dispatch";
-    });
-    assert.fail("expected the queued job to expire");
-  } catch (error) {
-    caught = error;
-  } finally {
-    releaseFirst();
-    await first;
-  }
-  return { caught, queuedDispatched };
-}
-
-test("#4165 a dispatched provider call is not killed by the queue budget", async () => {
-  const execution = await triggerQueueTimeout();
-  assert.equal(execution.dispatched, true, "the callback must enter execution");
-  assert.equal(execution.result, "should-not-reach");
-});
-
-test("#4165 queue expiry surfaces a clear local error", async () => {
-  const result = await triggerQueuedTimeout();
-  assert.ok(result.caught instanceof Error, "queue expiry must reject with an Error");
-  assert.equal(result.queuedDispatched, false, "an expired queued callback must never dispatch");
-  const caught = result.caught as Error & { code?: string };
-  assert.equal(caught.code, "RATE_LIMIT_QUEUE_TIMEOUT");
-  assert.match(caught.message, /maxWaitMs/);
-  assert.match(caught.message, /not an upstream/i);
-  assert.doesNotMatch(caught.message, /This job timed out/);
-});
-
-test("#4165 a job that completes within maxWaitMs is unaffected", async () => {
-  await rateLimitManager.applyRequestQueueSettings({
-    ...resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue,
-    autoEnableApiKeyProviders: false,
-    concurrentRequests: 1,
-    requestsPerMinute: 100000,
-    minTimeBetweenRequestsMs: 0,
-    maxWaitMs: 5000,
-  });
-  rateLimitManager.enableRateLimitProtection("conn-fast");
-
-  const result = await rateLimitManager.withRateLimit(
-    "openai",
-    "conn-fast",
-    "gpt-4o",
-    async () => "ok"
-  );
-  assert.equal(result, "ok");
-});
diff --git a/tests/unit/rate-limit-wedge-recovery.test.ts b/tests/unit/rate-limit-wedge-recovery.test.ts
index cd70a47934..7f9bfe865e 100644
--- a/tests/unit/rate-limit-wedge-recovery.test.ts
+++ b/tests/unit/rate-limit-wedge-recovery.test.ts
@@ -101,23 +101,3 @@ test("stop({ dropWaitingJobs: true }) rejects a genuinely queued (nothing-runnin
     "expected Bottleneck to reject with our dropErrorMessage verbatim"
   );
 });
-
-test("watchdog wedge branch uses stop({ dropWaitingJobs: true }), not disconnect()", async () => {
-  const source = await import("node:fs/promises").then((fs) =>
-    fs.readFile(new URL("../../open-sse/services/rateLimitManager.ts", import.meta.url), "utf8")
-  );
-
-  const wedgeBlockStart = source.indexOf("WEDGED:");
-  assert.ok(wedgeBlockStart >= 0, "expected to find the WEDGED log line in rateLimitManager.ts");
-  const wedgeBlock = source.slice(wedgeBlockStart, wedgeBlockStart + 1500);
-
-  assert.ok(
-    wedgeBlock.includes("stop({ dropWaitingJobs: true"),
-    "wedge-recovery branch must call stop({ dropWaitingJobs: true }) so orphaned queued jobs reject " +
-      "promptly instead of hanging until the outer per-target timeout (live incident 1784465227489-a2cbc0)"
-  );
-  assert.ok(
-    !/limiter\.disconnect\(\)/.test(wedgeBlock),
-    "wedge-recovery branch must not still call disconnect() — it doesn't reject queued jobs"
-  );
-});
diff --git a/tests/unit/ratelimit-admission-control-6593.test.ts b/tests/unit/ratelimit-admission-control-6593.test.ts
index 6eacf52696..521d6c6561 100644
--- a/tests/unit/ratelimit-admission-control-6593.test.ts
+++ b/tests/unit/ratelimit-admission-control-6593.test.ts
@@ -19,6 +19,7 @@ import os from "node:os";
 import path from "node:path";
 
 import { checkQueueAdmission } from "../../open-sse/services/rateLimitManager/admission.ts";
+import { getTrustedLocalRateLimitError } from "../../open-sse/services/rateLimitManager/errors.ts";
 
 const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-rl-admission-"));
 process.env.DATA_DIR = TEST_DATA_DIR;
@@ -78,6 +79,10 @@ test("#6593 checkQueueAdmission: rejects with a typed error at/over the cap", ()
   // also risks tripping the whole-provider circuit breaker for a purely local
   // admission decision.
   assert.equal(err?.status, 429);
+  assert.deepEqual(getTrustedLocalRateLimitError(err), {
+    code: "RATE_LIMIT_QUEUE_FULL",
+    status: 429,
+  });
   assert.match(err?.message ?? "", /maxQueueDepth/);
   assert.match(err?.message ?? "", /openai\/gpt-4o/);
 
@@ -106,16 +111,26 @@ test("#6593 withRateLimit: fast-fails once the queue is at the configured maxQue
   // Job 1 occupies the single concurrent slot. Poll (not a fixed sleep) until
   // Bottleneck has actually dispatched it, since QUEUED -> EXECUTING takes a
   // few event-loop ticks, not one.
-  const job1 = rateLimitManager.withRateLimit("openai", "conn-admission-cap", "gpt-4o", async () => {
-    await wait(150);
-    return "job1";
-  });
+  const job1 = rateLimitManager.withRateLimit(
+    "openai",
+    "conn-admission-cap",
+    "gpt-4o",
+    async () => {
+      await wait(150);
+      return "job1";
+    }
+  );
   await pollUntil(() => (status()?.executing ?? 0) + (status()?.running ?? 0) >= 1);
 
   // Job 2 has to wait behind job1 -> occupies the one allowed queue slot (QUEUED=1).
-  const job2 = rateLimitManager.withRateLimit("openai", "conn-admission-cap", "gpt-4o", async () => {
-    return "job2";
-  });
+  const job2 = rateLimitManager.withRateLimit(
+    "openai",
+    "conn-admission-cap",
+    "gpt-4o",
+    async () => {
+      return "job2";
+    }
+  );
   await pollUntil(() => (status()?.queued ?? 0) >= 1);
 
   // Job 3 arrives while QUEUED (1) is already at maxQueueDepth (1) -> fast-rejected.
@@ -124,6 +139,10 @@ test("#6593 withRateLimit: fast-fails once the queue is at the configured maxQue
     (err: Error & { code?: string; status?: number }) => {
       assert.equal(err.code, "RATE_LIMIT_QUEUE_FULL");
       assert.equal(err.status, 429);
+      assert.deepEqual(getTrustedLocalRateLimitError(err), {
+        code: "RATE_LIMIT_QUEUE_FULL",
+        status: 429,
+      });
       assert.match(err.message, /maxQueueDepth/);
       return true;
     }
@@ -161,10 +180,7 @@ test("#6593 withRateLimit: default maxQueueDepth=0 preserves unbounded-queue beh
 test("#6593 DEFAULT_REQUEST_QUEUE_MAX_WAIT_MS is 15s absent RATE_LIMIT_MAX_WAIT_MS", () => {
   assert.equal(process.env.RATE_LIMIT_MAX_WAIT_MS, undefined);
   assert.equal(resilienceSettings.DEFAULT_REQUEST_QUEUE_MAX_WAIT_MS, 15000);
-  assert.equal(
-    resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue.maxWaitMs,
-    15000
-  );
+  assert.equal(resilienceSettings.DEFAULT_RESILIENCE_SETTINGS.requestQueue.maxWaitMs, 15000);
 });
 
 test("#6593 DEFAULT_REQUEST_QUEUE_MAX_DEPTH defaults to 0 (disabled) absent an env override", () => {