diff --git a/extensions/azure-speech/speech-provider.test.ts b/extensions/azure-speech/speech-provider.test.ts
index fc847a171ec3..788d624dc20a 100644
--- a/extensions/azure-speech/speech-provider.test.ts
+++ b/extensions/azure-speech/speech-provider.test.ts
@@ -265,4 +265,32 @@ describe("buildAzureSpeechProvider", () => {
       timeoutMs: 45_000,
     });
   });
+
+  it("rejects blank credentials across readiness, discovery, and synthesis", async () => {
+    vi.stubEnv("AZURE_SPEECH_KEY", "   ");
+    vi.stubEnv("AZURE_SPEECH_API_KEY", "   ");
+    vi.stubEnv("SPEECH_KEY", "   ");
+    const provider = buildAzureSpeechProvider();
+    const providerConfig = { apiKey: "   ", region: "eastus" };
+
+    expect(provider.isConfigured({ providerConfig, timeoutMs: 1_000 })).toBe(false);
+    await expect(
+      provider.listVoices?.({ apiKey: "   ", providerConfig, timeoutMs: 1_000 }),
+    ).rejects.toThrow("Azure Speech API key missing");
+
+    const request = {
+      text: "hello",
+      cfg: {} as never,
+      providerConfig,
+      target: "audio-file" as const,
+      timeoutMs: 1_000,
+    };
+    await expect(provider.synthesize(request)).rejects.toThrow("Azure Speech API key missing");
+    await expect(provider.synthesizeTelephony?.(request)).rejects.toThrow(
+      "Azure Speech API key missing",
+    );
+
+    expect(listAzureSpeechVoicesMock).not.toHaveBeenCalled();
+    expect(azureSpeechTTSMock).not.toHaveBeenCalled();
+  });
 });
diff --git a/extensions/azure-speech/speech-provider.ts b/extensions/azure-speech/speech-provider.ts
index 6810da073dd9..495dfe63f173 100644
--- a/extensions/azure-speech/speech-provider.ts
+++ b/extensions/azure-speech/speech-provider.ts
@@ -9,7 +9,12 @@ import type {
   SpeechProviderOverrides,
   SpeechProviderPlugin,
 } from "openclaw/plugin-sdk/speech-core";
-import { asFiniteNumber, asObject, trimToUndefined } from "openclaw/plugin-sdk/speech-core";
+import {
+  asFiniteNumber,
+  asObject,
+  resolveSpeechProviderApiKey,
+  trimToUndefined,
+} from "openclaw/plugin-sdk/speech-core";
 import {
   azureSpeechTTS,
   DEFAULT_AZURE_SPEECH_AUDIO_FORMAT,
@@ -175,8 +180,8 @@ function parseDirectiveToken(ctx: SpeechDirectiveTokenParseContext): {
   }
 }
 
-function resolveApiKey(config: AzureSpeechProviderConfig): string | undefined {
-  return config.apiKey ?? readAzureSpeechEnvApiKey();
+function resolveApiKey(...candidates: Array<string | undefined>): string | undefined {
+  return resolveSpeechProviderApiKey(...candidates, readAzureSpeechEnvApiKey());
 }
 
 function resolveTimeoutMs(config: AzureSpeechProviderConfig, timeoutMs: number): number {
@@ -250,7 +255,9 @@ export function buildAzureSpeechProvider(): SpeechProviderPlugin {
       const config = req.providerConfig
         ? readAzureSpeechProviderConfig(req.providerConfig)
         : undefined;
-      const apiKey = req.apiKey ?? (config ? resolveApiKey(config) : readAzureSpeechEnvApiKey());
+      const requestValue = req.apiKey;
+      const configValue = config?.apiKey;
+      const apiKey = resolveApiKey(requestValue, configValue);
       if (!apiKey) {
         throw new Error("Azure Speech API key missing");
       }
@@ -264,12 +271,14 @@ export function buildAzureSpeechProvider(): SpeechProviderPlugin {
     },
     isConfigured: ({ providerConfig }) => {
       const config = readAzureSpeechProviderConfig(providerConfig);
-      return Boolean(resolveApiKey(config) && (config.baseUrl || config.region || config.endpoint));
+      return Boolean(
+        resolveApiKey(config.apiKey) && (config.baseUrl || config.region || config.endpoint),
+      );
     },
     synthesize: async (req) => {
       const config = readAzureSpeechProviderConfig(req.providerConfig);
       const overrides = readAzureSpeechOverrides(req.providerOverrides);
-      const apiKey = resolveApiKey(config);
+      const apiKey = resolveApiKey(config.apiKey);
       if (!apiKey) {
         throw new Error("Azure Speech API key missing");
       }
@@ -298,7 +307,7 @@ export function buildAzureSpeechProvider(): SpeechProviderPlugin {
     synthesizeTelephony: async (req) => {
       const config = readAzureSpeechProviderConfig(req.providerConfig);
       const overrides = readAzureSpeechOverrides(req.providerOverrides);
-      const apiKey = resolveApiKey(config);
+      const apiKey = resolveApiKey(config.apiKey);
       if (!apiKey) {
         throw new Error("Azure Speech API key missing");
       }
diff --git a/extensions/elevenlabs/speech-provider.test.ts b/extensions/elevenlabs/speech-provider.test.ts
index aee17f9d2652..5d3ed9af296f 100644
--- a/extensions/elevenlabs/speech-provider.test.ts
+++ b/extensions/elevenlabs/speech-provider.test.ts
@@ -5,6 +5,10 @@ import { buildElevenLabsSpeechProvider } from "./speech-provider.js";
 
 const fetchWithSsrFGuardMock = vi.hoisted(() => vi.fn());
 
+vi.mock("./config-api.js", () => ({
+  resolveElevenLabsApiKeyWithProfileFallback: () => null,
+}));
+
 vi.mock("openclaw/plugin-sdk/ssrf-runtime", () => ({
   fetchWithSsrFGuard: async (params: {
     url: string;
@@ -42,6 +46,7 @@ describe("elevenlabs speech provider", () => {
   afterEach(() => {
     globalThis.fetch = originalFetch;
     fetchWithSsrFGuardMock.mockClear();
+    vi.unstubAllEnvs();
     vi.restoreAllMocks();
   });
 
@@ -72,6 +77,35 @@ describe("elevenlabs speech provider", () => {
     );
   });
 
+  it("rejects blank credentials across discovery and synthesis before requests", async () => {
+    vi.stubEnv("ELEVENLABS_API_KEY", "   ");
+    vi.stubEnv("XI_API_KEY", "   ");
+    const provider = buildElevenLabsSpeechProvider();
+    const providerConfig = { apiKey: "   " };
+
+    expect(provider.isConfigured({ providerConfig, timeoutMs: 1_000 })).toBe(false);
+    await expect(
+      provider.listVoices?.({ apiKey: "   ", providerConfig, timeoutMs: 1_000 }),
+    ).rejects.toThrow("ElevenLabs API key missing");
+
+    const request = {
+      text: "hello",
+      cfg: {} as never,
+      providerConfig,
+      target: "audio-file" as const,
+      timeoutMs: 1_000,
+    };
+    await expect(provider.synthesize(request)).rejects.toThrow("ElevenLabs API key missing");
+    await expect(provider.streamSynthesize?.(request)).rejects.toThrow(
+      "ElevenLabs API key missing",
+    );
+    await expect(provider.synthesizeTelephony?.(request)).rejects.toThrow(
+      "ElevenLabs API key missing",
+    );
+
+    expect(fetchWithSsrFGuardMock).not.toHaveBeenCalled();
+  });
+
   it("keeps non-equivalent deprecated ElevenLabs TTS model IDs", async () => {
     const provider = buildElevenLabsSpeechProvider();
     const fetchMock = vi.fn(async (_url: string, init?: RequestInit) => {
diff --git a/extensions/elevenlabs/speech-provider.ts b/extensions/elevenlabs/speech-provider.ts
index 76c1d8e6bdf9..a1594083640d 100644
--- a/extensions/elevenlabs/speech-provider.ts
+++ b/extensions/elevenlabs/speech-provider.ts
@@ -24,6 +24,7 @@ import {
   requireInRange,
   trimToUndefined,
 } from "openclaw/plugin-sdk/speech";
+import { resolveSpeechProviderApiKey } from "openclaw/plugin-sdk/speech-core";
 import {
   fetchWithSsrFGuard,
   ssrfPolicyFromHttpBaseUrlAllowedHostname,
@@ -187,6 +188,24 @@ function readElevenLabsProviderConfig(config: SpeechProviderConfig): ElevenLabsP
   };
 }
 
+function resolveElevenLabsApiKey(...candidates: Array<string | undefined>): string | undefined {
+  return resolveSpeechProviderApiKey(
+    ...candidates,
+    resolveElevenLabsApiKeyWithProfileFallback() ?? undefined,
+    process.env.XI_API_KEY,
+  );
+}
+
+function resolveElevenLabsTalkApiKey(config: SpeechProviderConfig): string | undefined {
+  if (config.apiKey === undefined) {
+    return resolveElevenLabsApiKey();
+  }
+  return normalizeResolvedSecretInputString({
+    value: config.apiKey,
+    path: "talk.providers.elevenlabs.apiKey",
+  });
+}
+
 function mergeVoiceSettingsOverride(
   ctx: SpeechDirectiveTokenParseContext,
   next: Record<string, unknown>,
@@ -406,8 +425,7 @@ function resolveElevenLabsTtsRequest(
 ): Parameters<typeof elevenLabsTTS>[0] {
   const config = readElevenLabsProviderConfig(req.providerConfig);
   const overrides = req.providerOverrides ?? {};
-  const apiKey =
-    config.apiKey || resolveElevenLabsApiKeyWithProfileFallback() || process.env.XI_API_KEY;
+  const apiKey = resolveElevenLabsApiKey(config.apiKey);
   if (!apiKey) {
     throw new Error("ElevenLabs API key missing");
   }
@@ -441,13 +459,7 @@ export function buildElevenLabsSpeechProvider(): SpeechProviderPlugin {
     resolveTalkConfig: ({ baseTtsConfig, talkProviderConfig }) => {
       const base = normalizeElevenLabsProviderConfig(baseTtsConfig);
       const talkVoiceSettings = asObject(talkProviderConfig.voiceSettings);
-      const resolvedTalkApiKey =
-        talkProviderConfig.apiKey === undefined
-          ? (resolveElevenLabsApiKeyWithProfileFallback() ?? undefined)
-          : normalizeResolvedSecretInputString({
-              value: talkProviderConfig.apiKey,
-              path: "talk.providers.elevenlabs.apiKey",
-            });
+      const resolvedTalkApiKey = resolveElevenLabsTalkApiKey(talkProviderConfig);
       return {
         ...base,
         ...(resolvedTalkApiKey === undefined ? {} : { apiKey: resolvedTalkApiKey }),
@@ -529,11 +541,9 @@ export function buildElevenLabsSpeechProvider(): SpeechProviderPlugin {
       const config = req.providerConfig
         ? readElevenLabsProviderConfig(req.providerConfig)
         : undefined;
-      const apiKey =
-        req.apiKey ||
-        config?.apiKey ||
-        resolveElevenLabsApiKeyWithProfileFallback() ||
-        process.env.XI_API_KEY;
+      const requestValue = req.apiKey;
+      const configValue = config?.apiKey;
+      const apiKey = resolveElevenLabsApiKey(requestValue, configValue);
       if (!apiKey) {
         throw new Error("ElevenLabs API key missing");
       }
@@ -544,11 +554,7 @@ export function buildElevenLabsSpeechProvider(): SpeechProviderPlugin {
       });
     },
     isConfigured: ({ providerConfig }) =>
-      Boolean(
-        readElevenLabsProviderConfig(providerConfig).apiKey ||
-        resolveElevenLabsApiKeyWithProfileFallback() ||
-        process.env.XI_API_KEY,
-      ),
+      Boolean(resolveElevenLabsApiKey(readElevenLabsProviderConfig(providerConfig).apiKey)),
     synthesize: async (req) => {
       const overrides = req.providerOverrides ?? {};
       const outputFormat =
diff --git a/extensions/volcengine/speech-provider.ts b/extensions/volcengine/speech-provider.ts
index f519b42e0a5d..cb023b27bfd1 100644
--- a/extensions/volcengine/speech-provider.ts
+++ b/extensions/volcengine/speech-provider.ts
@@ -9,6 +9,7 @@ import type {
 import {
   asObject,
   parseSpeechDirectiveNumberOverride,
+  resolveSpeechProviderApiKey,
   trimToUndefined,
 } from "openclaw/plugin-sdk/speech-core";
 import { asFiniteNumberInRange } from "openclaw/plugin-sdk/string-coerce-runtime";
@@ -93,13 +94,23 @@ function normalizeVolcengineProviderConfig(
 }
 
 function resolveSeedSpeechApiKey(configApiKey?: string): string | undefined {
-  return (
-    configApiKey ??
-    trimToUndefined(process.env.VOLCENGINE_TTS_API_KEY) ??
-    trimToUndefined(process.env.BYTEPLUS_SEED_SPEECH_API_KEY)
+  return resolveSpeechProviderApiKey(
+    configApiKey,
+    process.env.VOLCENGINE_TTS_API_KEY,
+    process.env.BYTEPLUS_SEED_SPEECH_API_KEY,
   );
 }
 
+function resolveLegacyVolcengineCredentials(config: {
+  appId?: string;
+  token?: string;
+}): Pick<VolcengineTtsProviderConfig, "appId" | "token"> {
+  return {
+    appId: trimToUndefined(config.appId) ?? trimToUndefined(process.env.VOLCENGINE_TTS_APPID),
+    token: resolveSpeechProviderApiKey(config.token, process.env.VOLCENGINE_TTS_TOKEN),
+  };
+}
+
 function readProviderConfig(config: SpeechProviderConfig): VolcengineTtsProviderConfig {
   const normalized = normalizeVolcengineProviderConfig({});
   return {
@@ -187,19 +198,15 @@ export function buildVolcengineSpeechProvider(): SpeechProviderPlugin {
 
     isConfigured: ({ providerConfig }) => {
       const cfg = readProviderConfig(providerConfig);
-      return Boolean(
-        resolveSeedSpeechApiKey(cfg.apiKey) ||
-        ((cfg.appId || process.env.VOLCENGINE_TTS_APPID) &&
-          (cfg.token || process.env.VOLCENGINE_TTS_TOKEN)),
-      );
+      const legacy = resolveLegacyVolcengineCredentials(cfg);
+      return Boolean(resolveSeedSpeechApiKey(cfg.apiKey) || (legacy.appId && legacy.token));
     },
 
     synthesize: async (req) => {
       const cfg = readProviderConfig(req.providerConfig);
       const overrides = readVolcengineOverrides(req.providerOverrides);
       const apiKey = resolveSeedSpeechApiKey(cfg.apiKey);
-      const appId = cfg.appId || process.env.VOLCENGINE_TTS_APPID;
-      const token = cfg.token || process.env.VOLCENGINE_TTS_TOKEN;
+      const { appId, token } = resolveLegacyVolcengineCredentials(cfg);
 
       if (!apiKey && (!appId || !token)) {
         throw new Error(
diff --git a/extensions/volcengine/tts.test.ts b/extensions/volcengine/tts.test.ts
index af5c72efe288..8a616c936228 100644
--- a/extensions/volcengine/tts.test.ts
+++ b/extensions/volcengine/tts.test.ts
@@ -1,5 +1,5 @@
 // Volcengine tests cover tts plugin behavior.
-import { beforeEach, describe, expect, it, vi } from "vitest";
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
 import { buildVolcengineSpeechProvider } from "./speech-provider.js";
 import { volcengineTTS } from "./tts.js";
 
@@ -39,13 +39,12 @@ function makeLegacyProviderConfig(overrides?: Record<string, unknown>) {
   };
 }
 
-function clearTtsEnv() {
-  delete process.env.BYTEPLUS_API_KEY;
-  delete process.env.BYTEPLUS_SEED_SPEECH_API_KEY;
-  delete process.env.VOLCENGINE_TTS_API_KEY;
-  delete process.env.VOLCENGINE_TTS_APPID;
-  delete process.env.VOLCENGINE_TTS_TOKEN;
-}
+const TTS_ENV_KEYS = [
+  "BYTEPLUS_SEED_SPEECH_API_KEY",
+  "VOLCENGINE_TTS_API_KEY",
+  "VOLCENGINE_TTS_APPID",
+  "VOLCENGINE_TTS_TOKEN",
+] as const;
 
 function makeOversizedStreamResponse(): Response {
   return new Response(
@@ -59,19 +58,18 @@ function makeOversizedStreamResponse(): Response {
   );
 }
 
-function restoreOptionalEnv(key: string, value: string | undefined) {
-  if (value === undefined) {
-    delete process.env[key];
-  } else {
-    process.env[key] = value;
-  }
-}
-
 describe("Volcengine speech provider", () => {
   const provider = buildVolcengineSpeechProvider();
 
   beforeEach(() => {
     fetchWithSsrFGuardMock.mockReset();
+    for (const key of TTS_ENV_KEYS) {
+      vi.stubEnv(key, undefined);
+    }
+  });
+
+  afterEach(() => {
+    vi.unstubAllEnvs();
   });
 
   it("has correct id, label, and aliases", () => {
@@ -94,40 +92,32 @@ describe("Volcengine speech provider", () => {
   });
 
   it("reports not configured when credentials are missing", () => {
-    const oldBytePlusKey = process.env.BYTEPLUS_API_KEY;
-    const oldSeedKey = process.env.BYTEPLUS_SEED_SPEECH_API_KEY;
-    const oldApiKey = process.env.VOLCENGINE_TTS_API_KEY;
-    const oldAppId = process.env.VOLCENGINE_TTS_APPID;
-    const oldToken = process.env.VOLCENGINE_TTS_TOKEN;
-    clearTtsEnv();
-    try {
-      expect(provider.isConfigured({ providerConfig: {}, timeoutMs: 30000 })).toBe(false);
-    } finally {
-      restoreOptionalEnv("BYTEPLUS_API_KEY", oldBytePlusKey);
-      restoreOptionalEnv("BYTEPLUS_SEED_SPEECH_API_KEY", oldSeedKey);
-      restoreOptionalEnv("VOLCENGINE_TTS_API_KEY", oldApiKey);
-      restoreOptionalEnv("VOLCENGINE_TTS_APPID", oldAppId);
-      restoreOptionalEnv("VOLCENGINE_TTS_TOKEN", oldToken);
-    }
+    expect(provider.isConfigured({ providerConfig: {}, timeoutMs: 30000 })).toBe(false);
   });
 
   it("falls back to env vars for credentials", () => {
-    const oldBytePlusKey = process.env.BYTEPLUS_API_KEY;
-    const oldSeedKey = process.env.BYTEPLUS_SEED_SPEECH_API_KEY;
-    const oldApiKey = process.env.VOLCENGINE_TTS_API_KEY;
-    const oldAppId = process.env.VOLCENGINE_TTS_APPID;
-    const oldToken = process.env.VOLCENGINE_TTS_TOKEN;
-    clearTtsEnv();
-    process.env.BYTEPLUS_SEED_SPEECH_API_KEY = "env-api-key";
-    try {
-      expect(provider.isConfigured({ providerConfig: {}, timeoutMs: 30000 })).toBe(true);
-    } finally {
-      restoreOptionalEnv("BYTEPLUS_API_KEY", oldBytePlusKey);
-      restoreOptionalEnv("BYTEPLUS_SEED_SPEECH_API_KEY", oldSeedKey);
-      restoreOptionalEnv("VOLCENGINE_TTS_API_KEY", oldApiKey);
-      restoreOptionalEnv("VOLCENGINE_TTS_APPID", oldAppId);
-      restoreOptionalEnv("VOLCENGINE_TTS_TOKEN", oldToken);
+    vi.stubEnv("BYTEPLUS_SEED_SPEECH_API_KEY", "env-api-key");
+    expect(provider.isConfigured({ providerConfig: {}, timeoutMs: 30000 })).toBe(true);
+  });
+
+  it("rejects blank Seed and legacy credentials before requests", async () => {
+    for (const key of TTS_ENV_KEYS) {
+      vi.stubEnv(key, "   ");
     }
+    const providerConfig = { apiKey: "   ", appId: "   ", token: "   " };
+
+    expect(provider.isConfigured({ providerConfig, timeoutMs: 30_000 })).toBe(false);
+    await expect(
+      provider.synthesize({
+        text: "hello",
+        cfg: {},
+        providerConfig,
+        target: "audio-file",
+        timeoutMs: 1_000,
+      }),
+    ).rejects.toThrow("Volcengine TTS credentials missing");
+
+    expect(fetchWithSsrFGuardMock).not.toHaveBeenCalled();
   });
 
   it("lists voices with locale and gender", async () => {
diff --git a/extensions/xai/speech-provider.test.ts b/extensions/xai/speech-provider.test.ts
index 7e8baa6590a2..4c90799aec21 100644
--- a/extensions/xai/speech-provider.test.ts
+++ b/extensions/xai/speech-provider.test.ts
@@ -268,6 +268,29 @@ describe("xai speech provider", () => {
     });
   });
 
+  it("treats blank direct credentials as absent across readiness and requests", async () => {
+    process.env.XAI_API_KEY = "   ";
+    const provider = buildXaiSpeechProvider();
+    const providerConfig = { apiKey: "   " };
+
+    expect(provider.isConfigured({ cfg: {}, providerConfig, timeoutMs: 5_000 })).toBe(false);
+    await expect(provider.listVoices?.({ apiKey: "   ", providerConfig })).resolves.toEqual(
+      ["ara", "eve", "leo", "rex", "sal"].map((voice) => ({ id: voice, name: voice })),
+    );
+    await expect(
+      provider.synthesize({
+        text: "hello",
+        cfg: {},
+        providerConfig,
+        target: "audio-file",
+        timeoutMs: 5_000,
+      }),
+    ).rejects.toThrow("xAI credentials missing for TTS");
+
+    expect(listXaiTtsVoicesMock).not.toHaveBeenCalled();
+    expect(xaiTTSMock).not.toHaveBeenCalled();
+  });
+
   it("reports not configured when there is no apiKey, env, or auth profile", () => {
     isProviderAuthProfileConfiguredMock.mockReturnValue(false);
     const provider = buildXaiSpeechProvider();
diff --git a/extensions/xai/speech-provider.ts b/extensions/xai/speech-provider.ts
index 098431046b1b..f87165c3c846 100644
--- a/extensions/xai/speech-provider.ts
+++ b/extensions/xai/speech-provider.ts
@@ -13,6 +13,7 @@ import {
   type SpeechProviderPlugin,
   type SpeechSynthesisTarget,
 } from "openclaw/plugin-sdk/speech";
+import { resolveSpeechProviderApiKey } from "openclaw/plugin-sdk/speech-core";
 import {
   asFiniteNumberInRange,
   normalizeLowercaseStringOrEmpty,
@@ -136,6 +137,10 @@ function readXaiOverrides(overrides: SpeechProviderOverrides | undefined): XaiTt
   };
 }
 
+function resolveDirectXaiAudioApiKey(configApiKey?: string): string | undefined {
+  return resolveSpeechProviderApiKey(configApiKey, process.env.XAI_API_KEY);
+}
+
 function resolveGeneratedAudioMaxBytes(req: {
   cfg: { agents?: { defaults?: { mediaMaxMb?: number } } };
 }): number {
@@ -240,7 +245,7 @@ export function buildXaiSpeechProvider(): SpeechProviderPlugin {
       });
     },
     isConfigured: ({ providerConfig, cfg }) =>
-      Boolean(readXaiProviderConfig(providerConfig).apiKey || process.env.XAI_API_KEY) ||
+      Boolean(resolveDirectXaiAudioApiKey(readXaiProviderConfig(providerConfig).apiKey)) ||
       isProviderAuthProfileConfigured({ provider: "xai", cfg }),
     synthesize: async (req) => {
       const config = readXaiProviderConfig(req.providerConfig);
@@ -319,7 +324,7 @@ async function resolveOptionalXaiAudioApiKey(
   configApiKey: string | undefined,
   cfg?: OpenClawConfig,
 ): Promise<string | undefined> {
-  const direct = trimToUndefined(configApiKey) ?? trimToUndefined(process.env.XAI_API_KEY);
+  const direct = resolveDirectXaiAudioApiKey(configApiKey);
   if (direct) {
     return direct;
   }