@vellumai/assistant 0.10.12 → 0.11.0-staging.2

package/openapi.yaml

@@ -3,7 +3,7 @@
 openapi: 3.1.0
 info:
   title: Vellum Assistant API
-  version: 0.10.12
+  version: 0.11.0
   description: Auto-generated OpenAPI specification for the Vellum Assistant runtime HTTP server.
 servers:
   - url: http://127.0.0.1:7821
@@ -3999,6 +3999,7 @@ paths:
                             - email
                             - platform
                             - a2a
+                            - discord
                         label:
                           type: string
                         subtitle:
@@ -4146,6 +4147,34 @@ paths:
                       type: string
                     actorTeamId:
                       type: string
+                    appContext:
+                      type: object
+                      properties:
+                        entities:
+                          type: array
+                          items:
+                            type: object
+                            properties:
+                              type:
+                                type: string
+                              value:
+                                anyOf:
+                                  - type: string
+                                  - type: object
+                                    properties:
+                                      messageTs:
+                                        type: string
+                                      channelId:
+                                        type: string
+                              teamId:
+                                type: string
+                              enterpriseId:
+                                type: string
+                            required:
+                              - type
+                              - value
+                      required:
+                        - entities
                     admissionPolicy:
                       type: string
                       enum:
@@ -4282,6 +4311,7 @@ paths:
                             - email
                             - platform
                             - a2a
+                            - discord
                         ready:
                           type: boolean
                         setupStatus:
@@ -4371,6 +4401,7 @@ paths:
                     - email
                     - platform
                     - a2a
+                    - discord
                 includeRemote:
                   description: Include remote checks (default true)
                   type: boolean
@@ -4400,6 +4431,7 @@ paths:
                             - email
                             - platform
                             - a2a
+                            - discord
                         ready:
                           type: boolean
                         setupStatus:
@@ -6447,13 +6479,15 @@ paths:
           schema:
             type: string
             enum:
-              - slack
               - telegram
+              - phone
+              - vellum
               - whatsapp
+              - slack
               - email
+              - platform
               - a2a
-              - vellum
-              - phone
+              - discord
           description:
             Filter by origin channel. When provided, only conversations with this exact origin_channel value are
             returned. Omit to include all channels.
@@ -6579,6 +6613,7 @@ paths:
                                 - email
                                 - platform
                                 - a2a
+                                - discord
                             - type: "null"
                         assistantAttention:
                           type: object
@@ -6932,6 +6967,7 @@ paths:
                               - email
                               - platform
                               - a2a
+                              - discord
                           - type: "null"
                       assistantAttention:
                         type: object
@@ -8141,6 +8177,7 @@ paths:
                               - email
                               - platform
                               - a2a
+                              - discord
                           - type: "null"
                       assistantAttention:
                         type: object
@@ -14300,7 +14337,7 @@ paths:
             type: string
           description:
             "Filter by provider id. One of: anthropic, openai, gemini, ollama, fireworks, together, openrouter,
-            vercel-ai-gateway, litellm, openai-compatible, minimax, atlascloud, baseten"
+            vercel-ai-gateway, litellm, openai-compatible, minimax, atlascloud, baseten, poolside"
       responses:
         "200":
           description: Successful response
@@ -14531,7 +14568,7 @@ paths:
             type: string
           description:
             "Filter by provider. One of: anthropic, openai, gemini, ollama, fireworks, together, openrouter,
-            vercel-ai-gateway, litellm, openai-compatible, minimax, atlascloud, baseten, vellum"
+            vercel-ai-gateway, litellm, openai-compatible, minimax, atlascloud, baseten, poolside, vellum"
       responses:
         "200":
           description: Successful response
@@ -16357,6 +16394,192 @@ paths:
       responses:
         "200":
           description: Successful response
+  /v1/lifecycle/drain-status:
+    get:
+      operationId: lifecycle_drainstatus_get
+      summary: In-flight background work
+      description:
+        "Reports conversation turns, memory jobs, schedule runs, workflow runs, and heartbeat runs currently in
+        flight, plus the active quiesce lease. `idle: true` means it is safe to stop the assistant without interrupting
+        work."
+      tags:
+        - system
+      responses:
+        "200":
+          description: Successful response
+          content:
+            application/json:
+              schema:
+                type: object
+                properties:
+                  quiescedUntil:
+                    anyOf:
+                      - type: number
+                      - type: "null"
+                  idle:
+                    type: boolean
+                    description: True when no conversation turn, memory job, schedule run, workflow run, or heartbeat run is in flight.
+                  activeConversations:
+                    type: array
+                    items:
+                      type: object
+                      properties:
+                        conversationId:
+                          type: string
+                        title:
+                          anyOf:
+                            - type: string
+                            - type: "null"
+                        originChannel:
+                          anyOf:
+                            - type: string
+                            - type: "null"
+                        originInterface:
+                          anyOf:
+                            - type: string
+                            - type: "null"
+                        processingStartedAt:
+                          type: number
+                      required:
+                        - conversationId
+                        - title
+                        - originChannel
+                        - originInterface
+                        - processingStartedAt
+                      additionalProperties: false
+                  memoryJobs:
+                    type: array
+                    items:
+                      type: object
+                      properties:
+                        id:
+                          type: string
+                        type:
+                          type: string
+                        startedAt:
+                          anyOf:
+                            - type: number
+                            - type: "null"
+                      required:
+                        - id
+                        - type
+                        - startedAt
+                      additionalProperties: false
+                  scheduleRuns:
+                    type: array
+                    items:
+                      type: object
+                      properties:
+                        runId:
+                          type: string
+                        scheduleName:
+                          anyOf:
+                            - type: string
+                            - type: "null"
+                        startedAt:
+                          type: number
+                      required:
+                        - runId
+                        - scheduleName
+                        - startedAt
+                      additionalProperties: false
+                  workflowRuns:
+                    type: array
+                    items:
+                      type: object
+                      properties:
+                        runId:
+                          type: string
+                        name:
+                          anyOf:
+                            - type: string
+                            - type: "null"
+                        startedAt:
+                          anyOf:
+                            - type: number
+                            - type: "null"
+                      required:
+                        - runId
+                        - name
+                        - startedAt
+                      additionalProperties: false
+                  heartbeatRuns:
+                    type: array
+                    items:
+                      type: object
+                      properties:
+                        runId:
+                          type: string
+                        startedAt:
+                          type: number
+                      required:
+                        - runId
+                        - startedAt
+                      additionalProperties: false
+                required:
+                  - quiescedUntil
+                  - idle
+                  - activeConversations
+                  - memoryJobs
+                  - scheduleRuns
+                  - workflowRuns
+                  - heartbeatRuns
+                additionalProperties: false
+  /v1/lifecycle/quiesce:
+    delete:
+      operationId: lifecycle_quiesce_delete
+      summary: Release the drain quiesce lease
+      description:
+        Resumes background work immediately instead of waiting for the lease TTL to expire (e.g. when a drain is
+        cancelled).
+      tags:
+        - system
+      responses:
+        "200":
+          description: Successful response
+          content:
+            application/json:
+              schema:
+                type: object
+                properties:
+                  released:
+                    type: boolean
+                required:
+                  - released
+                additionalProperties: false
+    post:
+      operationId: lifecycle_quiesce_post
+      summary: Arm or refresh the drain quiesce lease
+      description:
+        Pauses the starting of new background work (heartbeat, schedules, watchers, sequences, memory jobs) for the
+        lease TTL. Refresh by calling again; the lease self-expires so an abandoned drain never leaves background work
+        paused.
+      tags:
+        - system
+      requestBody:
+        required: true
+        content:
+          application/json:
+            schema:
+              type: object
+              properties:
+                ttlMs:
+                  description: Lease TTL in ms (clamped to a sane range).
+                  type: number
+      responses:
+        "200":
+          description: Successful response
+          content:
+            application/json:
+              schema:
+                type: object
+                properties:
+                  quiescedUntil:
+                    type: number
+                    description: Epoch ms when the quiesce lease expires.
+                required:
+                  - quiescedUntil
+                additionalProperties: false
   /v1/live-voice/preflight:
     post:
       operationId: livevoice_preflight_post
@@ -19356,6 +19579,12 @@ paths:
                     - low
                     - medium
                     - high
+                bypassSecretCheck:
+                  description:
+                    When true, skip the secret-ingress scan for this message only. Set exclusively when the user explicitly
+                    confirms a client-side blocked send (the composer's "Send anyway" action); it is per-message and
+                    never persisted.
+                  type: boolean
                 hidden:
                   description:
                     "When true, persist the user message but suppress it from the UI transcript (it stays in LLM-side history
@@ -21176,6 +21405,7 @@ paths:
                     - email
                     - platform
                     - a2a
+                    - discord
                     - scheduler
                     - watcher
                 sourceContextId:
@@ -29104,7 +29334,9 @@ paths:
           required: false
           schema:
             type: string
-          description: Parent conversation ID (required)
+          description:
+            The subagent's own conversation ID. Fallback only — when the daemon knows the subagent (live or
+            rehydrated), it resolves the conversation itself and this parameter is ignored.
       responses:
         "200":
           description: Successful response
@@ -29167,6 +29399,10 @@ paths:
                         - type
                         - content
                       additionalProperties: false
+                  label:
+                    type: string
+                  parentToolUseId:
+                    type: string
                 required:
                   - subagentId
                   - events
@@ -29259,8 +29495,9 @@ paths:
       operationId: subagents_reconcile_get
       summary: Reconcile subagent live status
       description:
-        Returns the live in-memory status of all subagents known to the daemon for a given parent conversation.
-        Subagents not in the response are orphaned.
+        Returns the live in-memory status of all subagents known to the daemon for a given parent conversation,
+        with the identity fields a client needs to rebuild a store entry it never saw spawn (label, the subagent's own
+        conversation id, and the spawning tool-use id). Subagents not in the response are orphaned.
       tags:
         - subagents
       parameters:
@@ -29287,6 +29524,12 @@ paths:
                       properties:
                         status:
                           type: string
+                        label:
+                          type: string
+                        conversationId:
+                          type: string
+                        parentToolUseId:
+                          type: string
                       required:
                         - status
                       additionalProperties: false
@@ -31306,7 +31549,8 @@ paths:
       summary: Register a webhook callback URL
       description:
         Resolves a stable callback URL for a webhook type. On platform-managed assistants, registers the route with
-        the platform gateway. On self-hosted assistants, uses the configured ingress.publicBaseUrl.
+        the platform gateway. Otherwise uses the configured ingress.publicBaseUrl, falling back to the platform gateway
+        when no ingress is configured and the assistant is connected to the platform.
       tags:
         - webhooks
       requestBody:
@@ -32453,8 +32697,10 @@ components:
               anyOf:
                 - type: object
                   properties:
-                    mode:
-                      $ref: "#/components/schemas/ServiceMode"
+                    provider:
+                      type: string
+                    model:
+                      type: string
                   additionalProperties: {}
                 - type: "null"
             inference:
@@ -32483,6 +32729,7 @@ components:
         - together
         - litellm
         - baseten
+        - poolside
         - vellum
         - chatgpt
     ProfilePatchEntry:
@@ -32863,6 +33110,7 @@ components:
         - minimax
         - atlascloud
         - baseten
+        - poolside
         - vellum
     Auth:
       oneOf:
@@ -33265,8 +33513,10 @@ components:
             image-generation:
               type: object
               properties:
-                mode:
-                  $ref: "#/components/schemas/ServiceMode"
+                provider:
+                  type: string
+                model:
+                  type: string
               additionalProperties: {}
             inference:
               type: object

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vellumai/assistant",
-  "version": "0.10.12",
+  "version": "0.11.0-staging.2",
   "license": "MIT",
   "type": "module",
   "exports": {
@@ -42,7 +42,7 @@
     "@modelcontextprotocol/sdk": "1.27.1",
     "@qdrant/js-client-rest": "1.17.0",
     "@resvg/resvg-js": "2.6.2",
-    "@slack/types": "2.21.1",
+    "@slack/types": "2.22.0",
     "@vellumai/ces-client": "file:../packages/ces-client",
     "@vellumai/credential-storage": "file:../packages/credential-storage",
     "@vellumai/egress-proxy": "file:../packages/egress-proxy",

package/src/__tests__/acp-session.test.ts

@@ -2,7 +2,7 @@ import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 import { VellumAcpClientHandler } from "../acp/client-handler.js";
 import { AcpSessionManager } from "../acp/session-manager.js";
-import type { ServerMessage } from "../daemon/message-protocol.js";
+import type { AssistantEvent } from "../api/index.js";
 import * as pendingInteractions from "../runtime/pending-interactions.js";
 
 // ---------------------------------------------------------------------------
@@ -10,8 +10,8 @@ import * as pendingInteractions from "../runtime/pending-interactions.js";
 // ---------------------------------------------------------------------------
 
 describe("VellumAcpClientHandler", () => {
-  let sent: ServerMessage[];
-  let sendToVellum: (msg: ServerMessage) => void;
+  let sent: AssistantEvent[];
+  let sendToVellum: (msg: AssistantEvent) => void;
   let handler: VellumAcpClientHandler;
 
   beforeEach(() => {

package/src/__tests__/agent-loop-mutable-latest-user-message.test.ts

@@ -18,6 +18,9 @@ import { afterEach, describe, expect, mock, test } from "bun:test";
 let memoryV3LiveSlot = false;
 mock.module("../config/memory-v3-gate.js", () => ({
   isMemoryV3Live: () => memoryV3LiveSlot,
+  isV3TierActive: () => memoryV3LiveSlot,
+  isMemoryEnabled: (config?: { memory?: { enabled?: boolean } }) =>
+    config?.memory?.enabled !== false,
   usesConceptPageMemory: (memory?: {
     enabled?: boolean;
     v2?: { enabled?: boolean };

package/src/__tests__/anthropic-provider.test.ts

@@ -1915,6 +1915,46 @@ describe("AnthropicProvider — Cache-Control Characterization", () => {
     }
   });
 
+  test("a card-only assistant turn placeholders, and no card payload reaches the wire", async () => {
+    // `ui_surface` is a rendering instruction, never model context: the card's
+    // opaque payload must not reach the API, and the turn falls back to the
+    // sentinel to preserve role alternation. Conveying the card's meaning is
+    // the producer's job, via the `_surfaceFallback` text sibling — not
+    // something the provider reconstructs at request time.
+    const callSummaryBlock = {
+      type: "ui_surface",
+      surfaceId: "call-1",
+      surfaceType: "call_summary",
+      completed: true,
+      data: {
+        summaryText: "**Call completed** (42s). 3 event(s) recorded.",
+        status: "completed",
+        duration: 42,
+        events: [],
+      },
+    } as unknown as ContentBlock;
+    const messages: Message[] = [
+      userMsg("Start"),
+      { role: "assistant", content: [callSummaryBlock] },
+      userMsg("Continue"),
+    ];
+    await provider.sendMessage(messages);
+
+    const sent = lastStreamParams!.messages as Array<{
+      role: string;
+      content: Array<{ type: string; text?: string }>;
+    }>;
+
+    expect(sent).toHaveLength(3);
+    expect(sent[1].role).toBe("assistant");
+    expect(sent[1].content).toEqual([
+      { type: "text", text: PLACEHOLDER_BLOCKS_OMITTED },
+    ]);
+    // The opaque card payload itself never reaches the wire.
+    expect(JSON.stringify(sent)).not.toContain("call_summary");
+    expect(JSON.stringify(sent)).not.toContain("summaryText");
+  });
+
   test("assistant message with mix of known and unknown blocks keeps known blocks", async () => {
     const messages: Message[] = [
       userMsg("Start"),

package/src/__tests__/app-conversation-ids-backfill.test.ts

@@ -11,6 +11,7 @@ import {
 } from "../apps/app-store.js";
 import { getDb } from "../persistence/db-connection.js";
 import { initializeDb } from "../persistence/db-init.js";
+import { migrateBackfillAppConversationLineage } from "../persistence/migrations/354-backfill-app-conversation-lineage.js";
 import { rawRun } from "../persistence/raw-query.js";
 
 // Initialize db once for all tests
@@ -57,15 +58,18 @@ function insertMessage(
 }
 
 /** Insert a conversation row so FK constraints are satisfied. */
-function insertConversation(id: string): void {
+function insertConversation(id: string, parentConversationId?: string): void {
   const now = Date.now();
   rawRun(
     "test:insertConversation",
-    `INSERT OR IGNORE INTO conversations (id, title, created_at, updated_at) VALUES (?, ?, ?, ?)`,
+    `INSERT OR IGNORE INTO conversations (
+      id, title, created_at, updated_at, parent_conversation_id
+    ) VALUES (?, ?, ?, ?, ?)`,
     id,
     "test",
     now,
     now,
+    parentConversationId ?? null,
   );
 }
 
@@ -272,3 +276,30 @@ describe("backfillAppConversationIds", () => {
     expect(loaded?.conversationIds).toEqual([existingConvId, backfillConvId]);
   });
 });
+
+describe("migrateBackfillAppConversationLineage", () => {
+  test("adds inherited ancestors to existing direct app associations", () => {
+    const app = createApp(makeAppParams("Lineage App"));
+    const rootConversationId = "conv-lineage-root";
+    const parentConversationId = "conv-lineage-parent";
+    const childConversationId = "conv-lineage-child";
+    insertConversation(rootConversationId);
+    insertConversation(parentConversationId, rootConversationId);
+    insertConversation(childConversationId, parentConversationId);
+    addAppConversationId(app.id, childConversationId);
+
+    migrateBackfillAppConversationLineage();
+    migrateBackfillAppConversationLineage();
+
+    const loaded = getApp(app.id);
+    expect(loaded?.conversationIds).toEqual([
+      childConversationId,
+      parentConversationId,
+      rootConversationId,
+    ]);
+    expect(loaded?.inheritedConversationIds).toEqual([
+      parentConversationId,
+      rootConversationId,
+    ]);
+  });
+});