@vellumai/assistant 0.5.16 → 0.6.1

package/src/__tests__/task-runner.test.ts

@@ -1,4 +1,4 @@
-import { afterAll, beforeEach, describe, expect, mock, test } from "bun:test";
+import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 mock.module("../util/logger.js", () => ({
   getLogger: () =>
@@ -7,17 +7,13 @@ mock.module("../util/logger.js", () => ({
     }),
 }));
 
-import { getDb, initializeDb, resetDb } from "../memory/db.js";
+import { getDb, initializeDb } from "../memory/db.js";
 import { getTaskRunRules } from "../tasks/ephemeral-permissions.js";
 import { renderTemplate, runTask } from "../tasks/task-runner.js";
 import { createTask } from "../tasks/task-store.js";
 
 initializeDb();
 
-afterAll(() => {
-  resetDb();
-});
-
 // ── renderTemplate ──────────────────────────────────────────────────
 
 describe("renderTemplate", () => {

package/src/__tests__/task-scheduler.test.ts

@@ -1,4 +1,4 @@
-import { afterAll, beforeEach, describe, expect, mock, test } from "bun:test";
+import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 mock.module("../util/logger.js", () => ({
   getLogger: () =>
@@ -7,7 +7,7 @@ mock.module("../util/logger.js", () => ({
     }),
 }));
 
-import { getDb, initializeDb, resetDb } from "../memory/db.js";
+import { getDb, initializeDb } from "../memory/db.js";
 import {
   createSchedule,
   getSchedule,
@@ -33,10 +33,6 @@ function forceScheduleDue(scheduleId: string): void {
   ]);
 }
 
-afterAll(() => {
-  resetDb();
-});
-
 // ── scheduleTask helper ─────────────────────────────────────────────
 
 describe("scheduleTask", () => {

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

@@ -1,3 +1,4 @@
+import { existsSync, readFileSync } from "node:fs";
 import { afterEach, beforeEach, describe, expect, mock, test } from "bun:test";
 
 import type { ShellOutputResult } from "../tools/shared/shell-output.js";
@@ -66,7 +67,11 @@ mock.module("../tools/network/script-proxy/index.js", () => ({
 
 import type { SandboxConfig } from "../config/schema.js";
 import { parse } from "../tools/terminal/parser.js";
-import { buildSanitizedEnv } from "../tools/terminal/safe-env.js";
+import {
+  ALWAYS_INJECTED_ENV_VARS,
+  buildSanitizedEnv,
+  SAFE_ENV_VARS,
+} from "../tools/terminal/safe-env.js";
 import { wrapCommand } from "../tools/terminal/sandbox.js";
 import { ToolError } from "../util/errors.js";
 
@@ -450,29 +455,9 @@ describe("buildSanitizedEnv", () => {
   test("result is a plain object with no prototype-inherited secrets", () => {
     const env = buildSanitizedEnv();
     const keys = Object.keys(env);
-    const safeKeys = [
-      "PATH",
-      "HOME",
-      "TERM",
-      "LANG",
-      "EDITOR",
-      "SHELL",
-      "USER",
-      "TMPDIR",
-      "LC_ALL",
-      "LC_CTYPE",
-      "XDG_RUNTIME_DIR",
-      "DISPLAY",
-      "COLORTERM",
-      "TERM_PROGRAM",
-      "SSH_AUTH_SOCK",
-      "SSH_AGENT_PID",
-      "GPG_TTY",
-      "GNUPGHOME",
-      "VELLUM_DEV",
-      "INTERNAL_GATEWAY_BASE_URL",
-      "VELLUM_DATA_DIR",
-      "VELLUM_WORKSPACE_DIR",
+    const safeKeys: string[] = [
+      ...SAFE_ENV_VARS,
+      ...ALWAYS_INJECTED_ENV_VARS,
     ];
     for (const key of keys) {
       expect(safeKeys).toContain(key);
@@ -722,9 +707,14 @@ describe("formatShellOutput", () => {
   });
 
   test("truncates very long output", () => {
-    const longOutput = "x".repeat(60_000);
+    const longOutput = "x".repeat(30_000);
     const result = formatShellOutput(longOutput, "", 0, false, 120);
-    expect(result.content).toContain('<output_truncated limit="50K" />');
-    expect(result.content.length).toBeLessThan(60_000);
+    expect(result.content).toContain('limit="20K"');
+    // Extract the file="..." attribute from the truncation tag
+    const fileMatch = result.content.match(/file="([^"]+)"/);
+    expect(fileMatch).not.toBeNull();
+    const filePath = fileMatch![1];
+    expect(existsSync(filePath)).toBe(true);
+    expect(readFileSync(filePath, "utf-8")).toBe(longOutput);
   });
 });

package/src/__tests__/test-preload.ts

@@ -16,13 +16,20 @@ import { tmpdir } from "node:os";
 import { join } from "node:path";
 import { afterAll } from "bun:test";
 
+import { resetDb } from "../memory/db-connection.js";
+
 const testDir = realpathSync(
   mkdtempSync(join(tmpdir(), "vellum-test-workspace-")),
 );
 process.env.VELLUM_WORKSPACE_DIR = testDir;
+process.env.VELLUM_PLATFORM_URL = "https://test-platform.vellum.ai";
+process.exitCode = 0;
 
 afterAll(() => {
+  resetDb();
+  process.exitCode = 0;
   delete process.env.VELLUM_WORKSPACE_DIR;
+  delete process.env.VELLUM_PLATFORM_URL;
   try {
     rmSync(testDir, { recursive: true, force: true });
   } catch {

package/src/__tests__/tool-approval-handler.test.ts

@@ -1,4 +1,4 @@
-import { afterAll, beforeEach, describe, expect, mock, test } from "bun:test";
+import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 const testDir = process.env.VELLUM_WORKSPACE_DIR!;
 
@@ -43,7 +43,7 @@ import {
   mintGrantFromDecision,
   type MintGrantParams,
 } from "../approvals/approval-primitive.js";
-import { getDb, initializeDb, resetDb } from "../memory/db.js";
+import { getDb, initializeDb } from "../memory/db.js";
 import { scopedApprovalGrants } from "../memory/schema.js";
 import { computeToolApprovalDigest } from "../security/tool-approval-digest.js";
 import { ToolApprovalHandler } from "../tools/tool-approval-handler.js";
@@ -56,10 +56,6 @@ function clearTables(): void {
   db.delete(scopedApprovalGrants).run();
 }
 
-afterAll(() => {
-  resetDb();
-});
-
 // ---------------------------------------------------------------------------
 // Helpers
 // ---------------------------------------------------------------------------

package/src/__tests__/tool-executor.test.ts

@@ -2072,7 +2072,8 @@ describe("ToolExecutor persistentDecisionsAllowed contract", () => {
 // ---------------------------------------------------------------------------
 
 // Import the real buildSanitizedEnv (not mocked) for baseline credential tests
-const { buildSanitizedEnv } = await import("../tools/terminal/safe-env.js");
+const { buildSanitizedEnv, SAFE_ENV_VARS, ALWAYS_INJECTED_ENV_VARS } =
+  await import("../tools/terminal/safe-env.js");
 
 describe("buildSanitizedEnv — baseline: credential exclusion", () => {
   // Credential-like env vars that must never appear in the sanitized env.
@@ -2130,33 +2131,10 @@ describe("buildSanitizedEnv — baseline: credential exclusion", () => {
   });
 
   test("sanitized env only contains keys from the allowlist", () => {
-    const SAFE_ENV_VARS = [
-      "PATH",
-      "HOME",
-      "TERM",
-      "LANG",
-      "EDITOR",
-      "SHELL",
-      "USER",
-      "TMPDIR",
-      "LC_ALL",
-      "LC_CTYPE",
-      "XDG_RUNTIME_DIR",
-      "DISPLAY",
-      "COLORTERM",
-      "TERM_PROGRAM",
-      "SSH_AUTH_SOCK",
-      "SSH_AGENT_PID",
-      "GPG_TTY",
-      "GNUPGHOME",
-      "INTERNAL_GATEWAY_BASE_URL",
-      "VELLUM_DATA_DIR",
-      "VELLUM_WORKSPACE_DIR",
-    ];
-
+    const allowed: string[] = [...SAFE_ENV_VARS, ...ALWAYS_INJECTED_ENV_VARS];
     const env = buildSanitizedEnv();
     for (const key of Object.keys(env)) {
-      expect(SAFE_ENV_VARS).toContain(key);
+      expect(allowed).toContain(key);
     }
   });
 });

package/src/__tests__/tool-grant-request-escalation.test.ts

@@ -8,7 +8,7 @@
  * 5. Inline wait-and-resume for trusted-contact grant-gated tools
  */
 
-import { afterAll, beforeEach, describe, expect, mock, test } from "bun:test";
+import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 const testDir = process.env.VELLUM_WORKSPACE_DIR!;
 
@@ -109,7 +109,7 @@ import {
   getCanonicalGuardianRequest,
   listCanonicalGuardianRequests,
 } from "../memory/canonical-guardian-store.js";
-import { getDb, initializeDb, resetDb } from "../memory/db.js";
+import { getDb, initializeDb } from "../memory/db.js";
 import { scopedApprovalGrants } from "../memory/schema.js";
 import { computeToolApprovalDigest } from "../security/tool-approval-digest.js";
 import {
@@ -130,10 +130,6 @@ function resetTables(): void {
   db.run("DELETE FROM canonical_guardian_requests");
 }
 
-afterAll(() => {
-  resetDb();
-});
-
 // ---------------------------------------------------------------------------
 // Helpers
 // ---------------------------------------------------------------------------

package/src/__tests__/tool-side-effects-slack-dm.test.ts

@@ -0,0 +1,277 @@
+/**
+ * Tests for the bash post-execution hook that dispatches Slack DMs
+ * for channel verification sessions. Validates that the hook only
+ * delivers when an active session exists with a matching destination.
+ */
+
+import { beforeEach, describe, expect, mock, test } from "bun:test";
+
+// ---------------------------------------------------------------------------
+// Mocks — must be set up before importing the module under test
+// ---------------------------------------------------------------------------
+
+const mockFindActiveSession = mock(() => null as unknown);
+mock.module("../runtime/channel-verification-service.js", () => ({
+  findActiveSession: mockFindActiveSession,
+}));
+
+const mockDeliverVerificationSlack = mock(() => {});
+mock.module("../runtime/verification-outbound-actions.js", () => ({
+  deliverVerificationSlack: mockDeliverVerificationSlack,
+}));
+
+// Stub out transitive dependencies to prevent import errors
+mock.module("../bundler/app-compiler.js", () => ({
+  compileApp: mock(() => Promise.resolve({ ok: true })),
+}));
+mock.module("../media/app-icon-generator.js", () => ({
+  generateAppIcon: mock(() => Promise.resolve()),
+}));
+mock.module("../memory/app-store.js", () => ({
+  getApp: mock(() => null),
+  getAppDirPath: mock(() => ""),
+  isMultifileApp: mock(() => false),
+  resolveAppIdFromPath: mock(() => null),
+}));
+mock.module("../services/published-app-updater.js", () => ({
+  updatePublishedAppDeployment: mock(() => Promise.resolve()),
+}));
+mock.module("../daemon/conversation-surfaces.js", () => ({
+  refreshSurfacesForApp: mock(() => {}),
+}));
+mock.module("../daemon/doordash-steps.js", () => ({
+  isDoordashCommand: mock(() => false),
+  updateDoordashProgress: mock(() => {}),
+}));
+
+const mockLogWarn = mock((_obj: unknown, _msg: string) => {});
+const mockLogInfo = mock((_obj: unknown, _msg: string) => {});
+const mockLogError = mock((_obj: unknown, _msg: string) => {});
+mock.module("../util/logger.js", () => ({
+  getLogger: () => ({
+    warn: mockLogWarn,
+    info: mockLogInfo,
+    error: mockLogError,
+    debug: () => {},
+    trace: () => {},
+  }),
+}));
+
+// ---------------------------------------------------------------------------
+// Import after mocks — dynamic import ensures mock.module() calls above
+// are registered before tool-side-effects.ts evaluates its top-level
+// `const log = getLogger(...)`.
+// ---------------------------------------------------------------------------
+
+import type { SideEffectContext } from "../daemon/tool-side-effects.js";
+
+const { runPostExecutionSideEffects } =
+  await import("../daemon/tool-side-effects.js");
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+const dummySideEffectCtx = {
+  ctx: {} as SideEffectContext["ctx"],
+} satisfies SideEffectContext;
+
+function callWithBashHook(
+  command: string,
+  content: string,
+  isError = false,
+): void {
+  runPostExecutionSideEffects(
+    "bash",
+    { command },
+    { content, isError },
+    dummySideEffectCtx,
+  );
+}
+
+// ---------------------------------------------------------------------------
+// Tests
+// ---------------------------------------------------------------------------
+
+describe("bash hook — Slack DM dispatch with session validation", () => {
+  beforeEach(() => {
+    mockFindActiveSession.mockReset();
+    mockDeliverVerificationSlack.mockReset();
+    mockLogWarn.mockReset();
+    mockLogInfo.mockReset();
+    mockLogError.mockReset();
+  });
+
+  test("legitimate verification dispatches DM", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U123",
+      status: "awaiting_response",
+    });
+
+    const output = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U123",
+        text: "code",
+        assistantId: "aid",
+      },
+    });
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      output,
+    );
+
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledTimes(1);
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledWith(
+      "U123",
+      "code",
+      "aid",
+    );
+  });
+
+  test("no active session — DM not dispatched", () => {
+    mockFindActiveSession.mockReturnValue(null);
+
+    const output = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U123",
+        text: "code",
+        assistantId: "aid",
+      },
+    });
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      output,
+    );
+
+    expect(mockDeliverVerificationSlack).not.toHaveBeenCalled();
+    expect(mockLogWarn).toHaveBeenCalledWith(
+      expect.objectContaining({ userId: "U123" }),
+      expect.stringContaining("no active Slack verification session"),
+    );
+  });
+
+  test("userId mismatch — DM not dispatched", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U999",
+      status: "awaiting_response",
+    });
+
+    const output = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U_ATTACKER",
+        text: "code",
+        assistantId: "aid",
+      },
+    });
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      output,
+    );
+
+    expect(mockDeliverVerificationSlack).not.toHaveBeenCalled();
+    expect(mockLogWarn).toHaveBeenCalledWith(
+      expect.objectContaining({ userId: "U_ATTACKER", expected: "U999" }),
+      expect.stringContaining("does not match active session destination"),
+    );
+  });
+
+  test("command without verification substring — hook is no-op", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U123",
+      status: "awaiting_response",
+    });
+
+    const output = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U123",
+        text: "code",
+        assistantId: "aid",
+      },
+    });
+
+    callWithBashHook("echo hello", output);
+
+    expect(mockDeliverVerificationSlack).not.toHaveBeenCalled();
+  });
+
+  test("output without _pendingSlackDm — hook is no-op", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U123",
+      status: "awaiting_response",
+    });
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      JSON.stringify({ success: true, sessionId: "s1" }),
+    );
+
+    expect(mockDeliverVerificationSlack).not.toHaveBeenCalled();
+  });
+
+  test("multi-line JSON output — dispatches from correct line", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U123",
+      status: "awaiting_response",
+    });
+
+    const cancelResult = JSON.stringify({ success: true, cancelled: true });
+    const createResult = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U123",
+        text: "verify-code",
+        assistantId: "aid2",
+      },
+    });
+    const multiLineOutput = `${cancelResult}\n${createResult}`;
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      multiLineOutput,
+    );
+
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledTimes(1);
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledWith(
+      "U123",
+      "verify-code",
+      "aid2",
+    );
+  });
+
+  test("multi-line — rejected first line does not block valid second line", () => {
+    mockFindActiveSession.mockReturnValue({
+      destinationAddress: "U200",
+      status: "awaiting_response",
+    });
+
+    const staleResult = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U100",
+        text: "stale-code",
+        assistantId: "aid",
+      },
+    });
+    const validResult = JSON.stringify({
+      _pendingSlackDm: {
+        userId: "U200",
+        text: "valid-code",
+        assistantId: "aid2",
+      },
+    });
+    const multiLineOutput = `${staleResult}\n${validResult}`;
+
+    callWithBashHook(
+      "assistant channel-verification-sessions create ...",
+      multiLineOutput,
+    );
+
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledTimes(1);
+    expect(mockDeliverVerificationSlack).toHaveBeenCalledWith(
+      "U200",
+      "valid-code",
+      "aid2",
+    );
+  });
+});

package/src/__tests__/top-level-renderer.test.ts

@@ -15,11 +15,11 @@ describe("renderWorkspaceTopLevelContext", () => {
     const result = renderWorkspaceTopLevelContext(snapshot);
     expect(result).toBe(
       [
-        "<workspace_top_level>",
+        "<workspace>",
         "Root: /sandbox",
         "Directories: lib, src, tests",
         "Files: README.md, package.json",
-        "</workspace_top_level>",
+        "</workspace>",
       ].join("\n"),
     );
   });
@@ -61,11 +61,11 @@ describe("renderWorkspaceTopLevelContext", () => {
     const result = renderWorkspaceTopLevelContext(snapshot);
     expect(result).toBe(
       [
-        "<workspace_top_level>",
+        "<workspace>",
         "Root: /empty",
         "Directories: ",
         "Files: ",
-        "</workspace_top_level>",
+        "</workspace>",
       ].join("\n"),
     );
   });
@@ -92,8 +92,8 @@ describe("renderWorkspaceTopLevelContext", () => {
     };
 
     const result = renderWorkspaceTopLevelContext(snapshot);
-    expect(result.startsWith("<workspace_top_level>")).toBe(true);
-    expect(result.endsWith("</workspace_top_level>")).toBe(true);
+    expect(result.startsWith("<workspace>")).toBe(true);
+    expect(result.endsWith("</workspace>")).toBe(true);
   });
 
   test("includes hidden directories", () => {
@@ -124,7 +124,7 @@ describe("renderWorkspaceTopLevelContext", () => {
     expect(result).toContain("Files: a.txt, b.txt");
   });
 
-  test("renders current conversation and attachment paths when provided", () => {
+  test("renders attachment path when provided", () => {
     const snapshot: TopLevelSnapshot = {
       rootPath: "/sandbox",
       directories: ["src"],
@@ -133,16 +133,13 @@ describe("renderWorkspaceTopLevelContext", () => {
     };
 
     const result = renderWorkspaceTopLevelContext(snapshot, {
-      currentConversationPath: "conversations/2026-03-19T12-00-00.000Z_conv-1/",
-      currentConversationAttachmentsPath:
+      conversationAttachmentsPath:
         "conversations/2026-03-19T12-00-00.000Z_conv-1/attachments/",
     });
 
     expect(result).toContain(
-      "Current conversation folder: conversations/2026-03-19T12-00-00.000Z_conv-1/",
-    );
-    expect(result).toContain(
-      "Attachment files: conversations/2026-03-19T12-00-00.000Z_conv-1/attachments/",
+      "Current conversation attachments: conversations/2026-03-19T12-00-00.000Z_conv-1/attachments/",
     );
+    expect(result).not.toContain("Current conversation folder");
   });
 });

package/src/__tests__/trust-store.test.ts

@@ -771,7 +771,7 @@ describe("Trust Store", () => {
         "host_file_edit",
         "host_file_read",
         "host_file_write",
-        "memory_recall",
+        "recall",
         "scaffold_managed_skill",
         "skill_execute",
         "skill_load",

package/src/__tests__/trusted-contact-inline-approval-integration.test.ts

@@ -17,7 +17,7 @@
  *   f. Timeout/stale flow: guardian decision after prompt timeout produces deterministic outcome
  */
 
-import { afterAll, beforeEach, describe, expect, mock, test } from "bun:test";
+import { beforeEach, describe, expect, mock, test } from "bun:test";
 
 const testDir = process.env.VELLUM_WORKSPACE_DIR!;
 
@@ -169,7 +169,7 @@ import {
   listCanonicalGuardianRequests,
   updateCanonicalGuardianRequest,
 } from "../memory/canonical-guardian-store.js";
-import { getDb, initializeDb, resetDb } from "../memory/db.js";
+import { getDb, initializeDb } from "../memory/db.js";
 import { scopedApprovalGrants } from "../memory/schema.js";
 import { bridgeConfirmationRequestToGuardian } from "../runtime/confirmation-request-guardian-bridge.js";
 import { resolveRoutingState } from "../runtime/trust-context-resolver.js";
@@ -188,10 +188,6 @@ function resetTables(): void {
   db.run("DELETE FROM canonical_guardian_requests");
 }
 
-afterAll(() => {
-  resetDb();
-});
-
 // ---------------------------------------------------------------------------
 // Helpers
 // ---------------------------------------------------------------------------