fix(voice-call): harden inbound policy

extensions/voice-call/src/providers/telnyx.ts

@@ -21,15 +21,21 @@ import type { VoiceCallProvider } from "./base.js";
  * Uses Telnyx Call Control API v2 for managing calls.
  * @see https://developers.telnyx.com/docs/api/v2/call-control
  */
+export interface TelnyxProviderOptions {
+  /** Allow unsigned webhooks when no public key is configured */
+  allowUnsignedWebhooks?: boolean;
+}
+
 export class TelnyxProvider implements VoiceCallProvider {
   readonly name = "telnyx" as const;
 
   private readonly apiKey: string;
   private readonly connectionId: string;
   private readonly publicKey: string | undefined;
+  private readonly options: TelnyxProviderOptions;
   private readonly baseUrl = "https://api.telnyx.com/v2";
 
-  constructor(config: TelnyxConfig) {
+  constructor(config: TelnyxConfig, options: TelnyxProviderOptions = {}) {
     if (!config.apiKey) {
       throw new Error("Telnyx API key is required");
     }
@@ -40,6 +46,7 @@ export class TelnyxProvider implements VoiceCallProvider {
     this.apiKey = config.apiKey;
     this.connectionId = config.connectionId;
     this.publicKey = config.publicKey;
+    this.options = options;
   }
 
   /**
@@ -76,8 +83,14 @@ export class TelnyxProvider implements VoiceCallProvider {
    */
   verifyWebhook(ctx: WebhookContext): WebhookVerificationResult {
     if (!this.publicKey) {
-      // No public key configured, skip verification (not recommended for production)
-      return { ok: true };
+      if (this.options.allowUnsignedWebhooks) {
+        console.warn("[telnyx] Webhook verification skipped (no public key configured)");
+        return { ok: true, reason: "verification skipped (no public key configured)" };
+      }
+      return {
+        ok: false,
+        reason: "Missing telnyx.publicKey (configure to verify webhooks)",
+      };
     }
 
     const signature = ctx.headers["telnyx-signature-ed25519"];

extensions/voice-call/src/providers/twilio.test.ts

@@ -2,7 +2,7 @@ import { describe, expect, it } from "vitest";
 import type { WebhookContext } from "../types.js";
 import { TwilioProvider } from "./twilio.js";
 
-const STREAM_URL = "wss://example.ngrok.app/voice/stream";
+const STREAM_URL_PREFIX = "wss://example.ngrok.app/voice/stream?token=";
 
 function createProvider(): TwilioProvider {
   return new TwilioProvider(
@@ -24,13 +24,13 @@ function createContext(rawBody: string, query?: WebhookContext["query"]): Webhoo
 describe("TwilioProvider", () => {
   it("returns streaming TwiML for outbound conversation calls before in-progress", () => {
     const provider = createProvider();
-    const ctx = createContext("CallStatus=initiated&Direction=outbound-api", {
+    const ctx = createContext("CallStatus=initiated&Direction=outbound-api&CallSid=CA123", {
       callId: "call-1",
     });
 
     const result = provider.parseWebhookEvent(ctx);
 
-    expect(result.providerResponseBody).toContain(STREAM_URL);
+    expect(result.providerResponseBody).toContain(STREAM_URL_PREFIX);
     expect(result.providerResponseBody).toContain("<Connect>");
   });
 
@@ -50,11 +50,11 @@ describe("TwilioProvider", () => {
 
   it("returns streaming TwiML for inbound calls", () => {
     const provider = createProvider();
-    const ctx = createContext("CallStatus=ringing&Direction=inbound");
+    const ctx = createContext("CallStatus=ringing&Direction=inbound&CallSid=CA456");
 
     const result = provider.parseWebhookEvent(ctx);
 
-    expect(result.providerResponseBody).toContain(STREAM_URL);
+    expect(result.providerResponseBody).toContain(STREAM_URL_PREFIX);
     expect(result.providerResponseBody).toContain("<Connect>");
   });
 });

extensions/voice-call/src/providers/twilio.ts

@@ -60,6 +60,8 @@ export class TwilioProvider implements VoiceCallProvider {
 
   /** Map of call SID to stream SID for media streams */
   private callStreamMap = new Map<string, string>();
+  /** Per-call tokens for media stream authentication */
+  private streamAuthTokens = new Map<string, string>();
 
   /** Storage for TwiML content (for notify mode with URL-based TwiML) */
   private readonly twimlStorage = new Map<string, string>();
@@ -94,6 +96,7 @@ export class TwilioProvider implements VoiceCallProvider {
     }
 
     this.deleteStoredTwiml(callIdMatch[1]);
+    this.streamAuthTokens.delete(providerCallId);
   }
 
   constructor(config: TwilioConfig, options: TwilioProviderOptions = {}) {
@@ -138,6 +141,19 @@ export class TwilioProvider implements VoiceCallProvider {
     this.callStreamMap.delete(callSid);
   }
 
+  isValidStreamToken(callSid: string, token?: string): boolean {
+    const expected = this.streamAuthTokens.get(callSid);
+    if (!expected || !token) {
+      return false;
+    }
+    if (expected.length !== token.length) {
+      const dummy = Buffer.from(expected);
+      crypto.timingSafeEqual(dummy, dummy);
+      return false;
+    }
+    return crypto.timingSafeEqual(Buffer.from(expected), Buffer.from(token));
+  }
+
   /**
    * Clear TTS queue for a call (barge-in).
    * Used when user starts speaking to interrupt current TTS playback.
@@ -271,11 +287,13 @@ export class TwilioProvider implements VoiceCallProvider {
       case "busy":
       case "no-answer":
       case "failed":
+        this.streamAuthTokens.delete(callSid);
         if (callIdOverride) {
           this.deleteStoredTwiml(callIdOverride);
         }
         return { ...baseEvent, type: "call.ended", reason: callStatus };
       case "canceled":
+        this.streamAuthTokens.delete(callSid);
         if (callIdOverride) {
           this.deleteStoredTwiml(callIdOverride);
         }
@@ -308,6 +326,7 @@ export class TwilioProvider implements VoiceCallProvider {
     const callStatus = params.get("CallStatus");
     const direction = params.get("Direction");
     const isOutbound = direction?.startsWith("outbound") ?? false;
+    const callSid = params.get("CallSid") || undefined;
     const callIdFromQuery =
       typeof ctx.query?.callId === "string" && ctx.query.callId.trim()
         ? ctx.query.callId.trim()
@@ -330,7 +349,7 @@ export class TwilioProvider implements VoiceCallProvider {
 
       // Conversation mode: return streaming TwiML immediately for outbound calls.
       if (isOutbound) {
-        const streamUrl = this.getStreamUrl();
+        const streamUrl = callSid ? this.getStreamUrlForCall(callSid) : null;
         return streamUrl ? this.getStreamConnectXml(streamUrl) : TwilioProvider.PAUSE_TWIML;
       }
     }
@@ -343,7 +362,7 @@ export class TwilioProvider implements VoiceCallProvider {
     // Handle subsequent webhook requests (status callbacks, etc.)
     // For inbound calls, answer immediately with stream
     if (direction === "inbound") {
-      const streamUrl = this.getStreamUrl();
+      const streamUrl = callSid ? this.getStreamUrlForCall(callSid) : null;
       return streamUrl ? this.getStreamConnectXml(streamUrl) : TwilioProvider.PAUSE_TWIML;
     }
 
@@ -352,7 +371,7 @@ export class TwilioProvider implements VoiceCallProvider {
       return TwilioProvider.EMPTY_TWIML;
     }
 
-    const streamUrl = this.getStreamUrl();
+    const streamUrl = callSid ? this.getStreamUrlForCall(callSid) : null;
     return streamUrl ? this.getStreamConnectXml(streamUrl) : TwilioProvider.PAUSE_TWIML;
   }
 
@@ -380,6 +399,27 @@ export class TwilioProvider implements VoiceCallProvider {
     return `${wsOrigin}${path}`;
   }
 
+  private getStreamAuthToken(callSid: string): string {
+    const existing = this.streamAuthTokens.get(callSid);
+    if (existing) {
+      return existing;
+    }
+    const token = crypto.randomBytes(16).toString("base64url");
+    this.streamAuthTokens.set(callSid, token);
+    return token;
+  }
+
+  private getStreamUrlForCall(callSid: string): string | null {
+    const baseUrl = this.getStreamUrl();
+    if (!baseUrl) {
+      return null;
+    }
+    const token = this.getStreamAuthToken(callSid);
+    const url = new URL(baseUrl);
+    url.searchParams.set("token", token);
+    return url.toString();
+  }
+
   /**
    * Generate TwiML to connect a call to a WebSocket media stream.
    * This enables bidirectional audio streaming for real-time STT/TTS.
@@ -444,6 +484,7 @@ export class TwilioProvider implements VoiceCallProvider {
     this.deleteStoredTwimlForProviderCall(input.providerCallId);
 
     this.callWebhookUrls.delete(input.providerCallId);
+    this.streamAuthTokens.delete(input.providerCallId);
 
     await this.apiRequest(
       `/Calls/${input.providerCallId}.json`,