@vellumai/assistant 0.12.1-staging.1 → 0.12.1

package/src/__tests__/model-intents.test.ts

@@ -28,9 +28,6 @@ describe("model intents", () => {
     expect(resolveModelIntent("openai", "latency-optimized")).toBe(
       "gpt-5.6-luna",
     );
-    expect(resolveModelIntent("gemini", "cost-optimized")).toBe(
-      "gemini-3.1-flash-lite",
-    );
     expect(resolveModelIntent("gemini", "latency-optimized")).toBe(
       "gemini-3.1-flash-lite",
     );

package/src/__tests__/outlook-messaging-provider.test.ts

@@ -429,7 +429,7 @@ describe("Outlook messaging provider", () => {
   describe("sendMessage", () => {
     test("sends a new message with correct recipient, subject, and body", async () => {
       const conn = createMockConnection();
-      const result = await outlookMessagingProvider.sendMessage!(
+      const result = await outlookMessagingProvider.sendMessage(
         conn,
         "recipient@example.com",
         "Hello!",
@@ -452,7 +452,7 @@ describe("Outlook messaging provider", () => {
 
     test("uses empty string for subject when not provided", async () => {
       const conn = createMockConnection();
-      await outlookMessagingProvider.sendMessage!(
+      await outlookMessagingProvider.sendMessage(
         conn,
         "recipient@example.com",
         "Hello!",
@@ -471,7 +471,7 @@ describe("Outlook messaging provider", () => {
 
     test("calls replyToMessage when inReplyTo is provided", async () => {
       const conn = createMockConnection();
-      const result = await outlookMessagingProvider.sendMessage!(
+      const result = await outlookMessagingProvider.sendMessage(
         conn,
         "conv-id",
         "Reply text",
@@ -493,7 +493,7 @@ describe("Outlook messaging provider", () => {
 
     test("attaches files to a new message as Graph file attachments", async () => {
       const conn = createMockConnection();
-      await outlookMessagingProvider.sendMessage!(
+      await outlookMessagingProvider.sendMessage(
         conn,
         "recipient@example.com",
         "See attached",
@@ -530,7 +530,7 @@ describe("Outlook messaging provider", () => {
 
     test("forwards attachments to replyToMessage when replying", async () => {
       const conn = createMockConnection();
-      await outlookMessagingProvider.sendMessage!(
+      await outlookMessagingProvider.sendMessage(
         conn,
         "conv-id",
         "Reply body",
@@ -563,7 +563,7 @@ describe("Outlook messaging provider", () => {
 
     test("throws when connection is undefined", async () => {
       await expect(
-        outlookMessagingProvider.sendMessage!(
+        outlookMessagingProvider.sendMessage(
           undefined,
           "recipient@example.com",
           "Hello!",

package/src/__tests__/platform-client-verify-credential.test.ts

@@ -19,9 +19,7 @@ mock.module("../providers/platform-proxy/context.js", () => ({
   }),
 }));
 
-const actualEnv = await import("../config/env.js");
 mock.module("../config/env.js", () => ({
-  ...actualEnv,
   getPlatformAssistantId: () => platformAssistantId,
 }));
 

package/src/__tests__/plugin-import-boundary-guard.test.ts

@@ -129,9 +129,6 @@ const BASELINE: Record<string, readonly string[]> = {
     "../../../../daemon/embedding-reconcile.js",
     "../../../../daemon/trust-context.js",
     "../../../../daemon/turn-latency-sub-spans.js",
-    // mcp-setup capability cards list configured server names from the
-    // workspace mcp.json document. No plugin-api equivalent.
-    "../../../../mcp/workspace-mcp-config.js",
     "../../../../notifications/background-failure-signal.js",
     "../../../../persistence/checkpoints.js",
     "../../../../persistence/conversation-types.js",
@@ -213,16 +210,6 @@ const BASELINE: Record<string, readonly string[]> = {
     "../../../persistence/job-handlers/message-lexical.js",
     "../../../persistence/job-utils.js",
     "../../../persistence/jobs-store.js",
-    // The retrospective cursor is a `(createdAt, id)` bound that must keep
-    // working after its message row is deleted. The resolver lives beside
-    // the persistence after-reads it also drives, so the plugin's
-    // accounting imports it rather than re-deriving the bound. No
-    // plugin-api equivalent.
-    "../../../persistence/message-cursor.js",
-    // Named any-state `createdAt` lookup the cursor bookkeeping uses for
-    // forks and the timestamp backfill; the messages-read boundary guard
-    // requires the read to live in persistence. No plugin-api equivalent.
-    "../../../persistence/message-reads.js",
     "../../../persistence/raw-query.js",
     "../../../persistence/schema/index.js",
     "../../../prompts/persona-resolver.js",

package/src/__tests__/script-proxy-certs.test.ts

@@ -8,7 +8,6 @@ import {
   getCAPath,
   issueLeafCert,
 } from "../outbound-proxy/index.js";
-import { writeCombinedCABundle } from "../util/ca-bundle.js";
 
 let dataDir: string;
 
@@ -20,40 +19,6 @@ afterAll(async () => {
   await rm(dataDir, { recursive: true, force: true });
 });
 
-test("combined trust verifies certificates from both CA sources", async () => {
-  const roots = [join(dataDir, "system"), join(dataDir, "extra")];
-  const leaves: string[] = [];
-  for (const root of roots) {
-    await ensureLocalCA(root);
-    const caDir = join(root, "proxy-ca");
-    await issueLeafCert(caDir, "example.com");
-    leaves.push(join(caDir, "issued", "example.com.pem"));
-  }
-  const bundle = join(dataDir, "combined.pem");
-  await writeCombinedCABundle(
-    getCAPath(roots[0]!),
-    getCAPath(roots[1]!),
-    bundle,
-  );
-  const verify = Bun.spawn(
-    ["openssl", "verify", "-CAfile", bundle, ...leaves],
-    {
-      stdout: "pipe",
-      stderr: "pipe",
-      windowsHide: true,
-    },
-  );
-  const [code, output, error] = await Promise.all([
-    verify.exited,
-    new Response(verify.stdout).text(),
-    new Response(verify.stderr).text(),
-  ]);
-  expect({ code, error }).toEqual({ code: 0, error: "" });
-  for (const leaf of leaves) {
-    expect(output).toContain(`${leaf}: OK`);
-  }
-});
-
 describe("ensureLocalCA", () => {
   test("creates CA cert and key with correct permissions", async () => {
     await ensureLocalCA(dataDir);

package/src/__tests__/skill-feature-flags-integration.test.ts

@@ -60,7 +60,7 @@ function makeConfig(overrides: Partial<AssistantConfig> = {}): AssistantConfig {
   return {
     skills: {
       entries: {},
-      load: { extraDirs: [] },
+      load: { extraDirs: [], watch: true, watchDebounceMs: 250 },
       install: { nodeManager: "npm" },
       allowBundled: null,
       remoteProviders: {

package/src/__tests__/skill-feature-flags.test.ts

@@ -28,7 +28,7 @@ function makeConfig(overrides: Partial<AssistantConfig> = {}): AssistantConfig {
   return {
     skills: {
       entries: {},
-      load: { extraDirs: [] },
+      load: { extraDirs: [], watch: true, watchDebounceMs: 250 },
       install: { nodeManager: "npm" },
       allowBundled: null,
       remoteProviders: {
@@ -233,7 +233,7 @@ describe("resolveSkillStates with feature flags", () => {
     const config = makeConfig({
       skills: {
         entries: { [DECLARED_SKILL_ID]: { enabled: true } },
-        load: { extraDirs: [] },
+        load: { extraDirs: [], watch: true, watchDebounceMs: 250 },
         install: { nodeManager: "npm" },
         allowBundled: null,
         remoteProviders: {

package/src/__tests__/speaker-identification.test.ts

@@ -0,0 +1,56 @@
+import { describe, expect, test } from "bun:test";
+
+import {
+  extractPromptSpeakerMetadata,
+  SpeakerIdentityTracker,
+} from "../calls/speaker-identification.js";
+
+describe("speaker-identification", () => {
+  test("extractPromptSpeakerMetadata: reads top-level and nested fields", () => {
+    const metadata = extractPromptSpeakerMetadata({
+      speaker_id: "spk-7",
+      speaker_label: "Conference room mic",
+      speaker_confidence: "0.88",
+      metadata: {
+        participantId: "participant-22",
+      },
+    });
+
+    expect(metadata.speakerId).toBe("spk-7");
+    expect(metadata.speakerLabel).toBe("Conference room mic");
+    expect(metadata.speakerConfidence).toBe(0.88);
+    expect(metadata.participantId).toBe("participant-22");
+  });
+
+  test("SpeakerIdentityTracker: keeps stable identity for provider speaker id", () => {
+    const tracker = new SpeakerIdentityTracker();
+
+    const first = tracker.identifySpeaker({
+      speakerId: "speaker-a",
+      speakerName: "Aaron",
+      speakerConfidence: 0.93,
+    });
+    const second = tracker.identifySpeaker({
+      speakerId: "speaker-a",
+      speakerName: "Aaron",
+      speakerConfidence: 0.81,
+    });
+
+    expect(first.speakerId).toBe("speaker-a");
+    expect(first.speakerLabel).toBe("Aaron");
+    expect(first.source).toBe("provider");
+    expect(second.speakerId).toBe("speaker-a");
+    expect(second.speakerLabel).toBe("Aaron");
+    expect(second.speakerConfidence).toBe(0.81);
+    expect(tracker.listProfiles()).toHaveLength(1);
+  });
+
+  test("SpeakerIdentityTracker: falls back to inferred primary speaker without provider ids", () => {
+    const tracker = new SpeakerIdentityTracker();
+    const speaker = tracker.identifySpeaker({});
+
+    expect(speaker.speakerId).toBe("primary-speaker");
+    expect(speaker.speakerLabel).toBe("Speaker 1");
+    expect(speaker.source).toBe("inferred");
+  });
+});

package/src/__tests__/subagent-tool-gate-mode.test.ts

@@ -249,23 +249,14 @@ describe("createResolveToolsCallback — toolContextPin", () => {
     });
   }
 
-  test("control: without a pin, a clientless fork drops client-gated tools but keeps host and UI tools", () => {
+  test("control: without a pin, a clientless fork drops every client-gated tool from the wire", () => {
     projectedSkillToolNames = [];
     const resolve = createResolveToolsCallback(
       CLIENT_GATED_DEFS,
       clientlessExecutionCtx(),
     )!;
 
-    // Host tool definitions stay on the wire for background turns, and
-    // ui_show stays because background UI surfaces persist and return. The
-    // executor rejects host tools on explicitly non-interactive turns before
-    // dispatch; ask_question and request_system_permission remain
-    // client-gated at resolution.
-    expect(resolve(EMPTY_HISTORY).map((t) => t.name)).toEqual([
-      "remember",
-      "host_bash",
-      "ui_show",
-    ]);
+    expect(resolve(EMPTY_HISTORY).map((t) => t.name)).toEqual(["remember"]);
   });
 
   test("a desktop-source pin restores the host/UI/client tool defs on the wire", () => {
@@ -300,14 +291,7 @@ describe("createResolveToolsCallback — toolContextPin", () => {
       }),
     )!;
 
-    // Host tool definitions stay on the wire for background turns, and
-    // ui_show survives the clientless pin: it persists and returns. The
-    // executor rejects host tools on explicitly non-interactive turns.
-    expect(resolve(EMPTY_HISTORY).map((t) => t.name)).toEqual([
-      "remember",
-      "host_bash",
-      "ui_show",
-    ]);
+    expect(resolve(EMPTY_HISTORY).map((t) => t.name)).toEqual(["remember"]);
   });
 
   test("invariant: a pinned-in tool is on the wire but can never execute", async () => {

package/src/__tests__/subagent-tools.test.ts

@@ -3447,9 +3447,8 @@ describe("Subagent advisor-role consult", () => {
   });
 
   test("advisor consults still run when the parent conversation is gone", async () => {
-    // Advisor consults do not look up the parent conversation for a skill
-    // catalog snapshot. An evicted parent still gets a consult; the context
-    // pack loads the catalog from disk independently.
+    // The parent is looked up only for its warm skill catalog, so an evicted
+    // conversation costs a section of the context pack, not the consult.
     mockFindConversation = () => undefined;
     const { captured, restore } = stubSpawn();
     try {

package/src/__tests__/telephony-tts-guard.test.ts

@@ -455,11 +455,10 @@ describe("resolveCallTtsProvider with requiresPcmAudio", () => {
     expect(result.provider?.id).toBe("elevenlabs");
   });
 
-  test("without requiresPcmAudio the configured provider is not swapped for playability", async () => {
+  test("without requiresPcmAudio the configured provider is not swapped", async () => {
     testConfig.services.tts.provider = "deepgram";
-    storedKeys["deepgram"] = "dg-key";
-    // The CR transport path does not consult the media-stream playability
-    // capability.
+    // No credentials at all — the CR transport path does not consult the
+    // media-stream playability capability.
     registerStubProvider("deepgram", async () => {
       return { audio: Buffer.from("x"), contentType: "audio/mpeg" };
     });
@@ -469,37 +468,6 @@ describe("resolveCallTtsProvider with requiresPcmAudio", () => {
     expect(result.useSynthesizedPath).toBe(true);
     expect(result.audioFormat).toBe("mp3");
   });
-
-  test("managed speech stands in for a configured provider whose credential is missing", async () => {
-    testConfig.services.tts.provider = "deepgram";
-    // No deepgram key, managed speech available: the phone speaks through
-    // managed speech, the same substitution live voice makes.
-    registerStubProvider("deepgram", async () => {
-      throw new Error("should not be used");
-    });
-    registerStubProvider("vellum", async () => {
-      return { audio: Buffer.from("x"), contentType: "audio/mpeg" };
-    });
-
-    const result = await resolveCallTtsProvider();
-    expect(result.provider?.id).toBe("vellum");
-    expect(result.useSynthesizedPath).toBe(true);
-  });
-
-  test("without managed speech a credential-less configured provider is kept", async () => {
-    mockManagedSpeechAvailable = false;
-    try {
-      testConfig.services.tts.provider = "deepgram";
-      registerStubProvider("deepgram", async () => {
-        return { audio: Buffer.from("x"), contentType: "audio/mpeg" };
-      });
-
-      const result = await resolveCallTtsProvider();
-      expect(result.provider?.id).toBe("deepgram");
-    } finally {
-      mockManagedSpeechAvailable = true;
-    }
-  });
 });
 
 // ---------------------------------------------------------------------------
@@ -616,7 +584,6 @@ describe("speakSystemPrompt mid-stream failure after audio started", () => {
   test("non-PCM transport: skips the native text fallback — no full-prompt re-speak", async () => {
     testConfig.services.tts.provider = "fish-audio";
     testConfig.services.tts.providers["fish-audio"].referenceId = "ref-123";
-    storedKeys["fish-audio"] = "fa-key";
 
     const fishStream = jest.fn(failAfterFirstChunk());
     registerStreamingStubProvider("fish-audio", fishStream);
@@ -690,7 +657,6 @@ describe("speakSystemPrompt aborted synthesis", () => {
   test("non-PCM transport with native-fallback provider: abort skips the text fallback and end-of-turn", async () => {
     testConfig.services.tts.provider = "fish-audio";
     testConfig.services.tts.providers["fish-audio"].referenceId = "ref-123";
-    storedKeys["fish-audio"] = "fa-key";
 
     const controller = new AbortController();
     controller.abort();
@@ -776,7 +742,6 @@ describe("speakSystemPrompt aborted synthesis", () => {
   test("provider AbortError without any signal is a failure — non-PCM native fallback still runs", async () => {
     testConfig.services.tts.provider = "fish-audio";
     testConfig.services.tts.providers["fish-audio"].referenceId = "ref-123";
-    storedKeys["fish-audio"] = "fa-key";
 
     const fishSynthesize = jest.fn(async () => {
       throw new DOMException("socket torn down", "AbortError");

package/src/__tests__/vellum-self-knowledge-inline-command.test.ts

@@ -136,10 +136,6 @@ describe("vellum-self-knowledge skill", () => {
     const result = await executeSkillLoad({ skill: "vellum-self-knowledge" });
     expect(result.content).toContain("## Critical Rule");
     expect(result.content).toContain("Never answer from memory");
-    expect(result.content).toContain("distinct, legacy Vellum product");
-    expect(result.content).toContain(
-      "Never use it to answer questions about this Vellum",
-    );
   });
 
   // ── Resolution order ─────────────────────────────────────────────────

package/src/{calls/__tests__/voice-session-bridge-turn-flow.test.ts → __tests__/voice-session-bridge.test.ts}

@@ -1,21 +1,13 @@
-/**
- * Tests for `startVoiceTurn`'s turn flow against real persistence: delta and
- * error forwarding, abort handling, persisted channel metadata, the auto-built
- * phone control prompt, and the approval-policy split between guardian and
- * non-guardian callers. The sibling `voice-session-bridge.test.ts` covers the
- * routing legs, transcript hygiene, tool-event forwarding, and the lock and
- * teardown races against scripted doubles.
- */
 import { beforeEach, describe, expect, mock, test } from "bun:test";
 
-import { setConfig } from "../../__tests__/helpers/set-config.js";
-import type { AssistantEvent } from "../../api/index.js";
+import type { AssistantEvent } from "../api/index.js";
 import type {
   TurnChannelContext,
   TurnInterfaceContext,
-} from "../../channels/types.js";
-import type { Conversation } from "../../daemon/conversation.js";
-import { persistUserMessage as persistUserMessageImpl } from "../../daemon/conversation-messaging.js";
+} from "../channels/types.js";
+import type { Conversation } from "../daemon/conversation.js";
+import { persistUserMessage as persistUserMessageImpl } from "../daemon/conversation-messaging.js";
+import { setConfig } from "./helpers/set-config.js";
 
 /** Seed the config the voice bridge reads: disclosure copy, plus disabled
  * secret detection and memory so the real persist path stays inert. */
@@ -27,7 +19,7 @@ function seedVoiceConfig(disclosure: { enabled: boolean; text: string }): void {
 
 let voiceConversationFactory: (() => Conversation) | null = null;
 
-mock.module("../../daemon/conversation-store.js", () => ({
+mock.module("../daemon/conversation-store.js", () => ({
   getOrCreateConversation: async () => {
     if (!voiceConversationFactory) {
       throw new Error("voiceConversationFactory not set for test");
@@ -36,18 +28,18 @@ mock.module("../../daemon/conversation-store.js", () => ({
   },
 }));
 
+import { CALL_OPENING_MARKER } from "../calls/voice-control-protocol.js";
+import { startVoiceTurn } from "../calls/voice-session-bridge.js";
 import {
   createConversation,
   getMessages,
-} from "../../persistence/conversation-crud.js";
-import { getDb } from "../../persistence/db-connection.js";
-import { initializeDb } from "../../persistence/db-init.js";
+} from "../persistence/conversation-crud.js";
+import { getDb } from "../persistence/db-connection.js";
+import { initializeDb } from "../persistence/db-init.js";
 import {
   assistantEventHub,
   broadcastMessage,
-} from "../../runtime/assistant-event-hub.js";
-import { CALL_OPENING_MARKER } from "../voice-control-protocol.js";
-import { startVoiceTurn } from "../voice-session-bridge.js";
+} from "../runtime/assistant-event-hub.js";
 
 await initializeDb();
 

package/src/__tests__/watch-retro-tool-availability.test.ts

@@ -1,10 +1,13 @@
 /**
  * The gate that decides whether a watch retrospective can report at all.
  *
- * A retrospective runs as a `clientless` wake, but its report continues to
- * use the post-turn renderer so the report lands after its tool call has been
- * persisted. The core UI tools remain available on that turn for other
- * background work, so this file pins both contracts against the real registry.
+ * A retrospective runs as a `clientless` wake, which pins the turn
+ * non-interactive. `conversation-tool-setup` gates the whole `ui_surface`
+ * family on a client being present, so in that turn `ui_show` is not denied,
+ * it is absent: a retrospective told to call it can only tell the user it
+ * cannot. Nothing about the card's schema, its renderer, or the prompt's
+ * wording reveals that, which is why it is asserted here against the real
+ * registry rather than assumed anywhere else.
  */
 
 import { afterAll, describe, expect, test } from "bun:test";
@@ -51,14 +54,20 @@ describe("watch retrospective tool availability", () => {
     ).toBe(true);
   });
 
-  test("core UI tools stay available on a clientless turn", async () => {
+  test("ui_show is absent from a clientless turn", async () => {
     await initializeTools();
     const ctx = clientlessContext();
 
-    for (const name of ["ui_show", "ui_update", "ui_dismiss"]) {
-      expect(isToolActiveForContext(name, ctx)).toBe(true);
-    }
+    // The reason the report does not go through `ui_show`. If this ever flips
+    // to true, the extra tool above can be reconsidered; while it is false,
+    // routing the retrospective's card through `ui_show` produces a turn that
+    // cannot report at all.
+    expect(isToolActiveForContext("ui_show", ctx)).toBe(false);
+    expect(isToolActiveForContext("ui_update", ctx)).toBe(false);
+    expect(isToolActiveForContext("ui_dismiss", ctx)).toBe(false);
 
+    // And the gate really is about the client, not about the tools being
+    // unregistered: with one attached, the same names are active.
     const withClient = clientfulContext();
     expect(isToolActiveForContext("ui_show", withClient)).toBe(true);
   });

package/src/__tests__/workspace-git-service.test.ts

@@ -4,7 +4,6 @@ import {
   existsSync,
   mkdirSync,
   readFileSync,
-  renameSync,
   rmSync,
   statSync,
   symlinkSync,
@@ -524,31 +523,6 @@ describe("WorkspaceGitService", () => {
       expect(trackedFiles()).not.toContain("gone-b.txt");
       expect(trackedFiles()).not.toContain("gone-c.txt");
     });
-
-    test("batched rename stages origin and destination even when they split", async () => {
-      const service = new WorkspaceGitService(testDir);
-      await service.ensureInitialized();
-
-      writeFileSync(join(testDir, "old-name.txt"), "moved");
-      await service.commitChanges("Seed rename source");
-
-      renameSync(join(testDir, "old-name.txt"), join(testDir, "new-name.txt"));
-      for (const name of ["x0.txt", "x1.txt", "x2.txt", "x3.txt"]) {
-        writeFileSync(join(testDir, name), name);
-      }
-
-      await withStageBatchSize(2, async () => {
-        await service.commitChanges("Rename across batches");
-      });
-
-      expect(trackedFiles()).toContain("new-name.txt");
-      expect(trackedFiles()).not.toContain("old-name.txt");
-      const status = execFileSync("git", ["status", "--porcelain"], {
-        cwd: testDir,
-        encoding: "utf-8",
-      }).trim();
-      expect(status).toBe("");
-    });
   });
 
   describe("getStatus", () => {