@checkstack/backend 0.9.0 → 0.9.1

package/CHANGELOG.md

@@ -1,5 +1,135 @@
 # @checkstack/backend
 
+## 0.9.1
+
+### Patch Changes
+
+- aa89bc5: Replace the bespoke `registerInfrastructureTab()` registry with a standard
+  slot-extension contract (`InfrastructureTabsSlot` from
+  `@checkstack/infrastructure-common`). Plugins now contribute infrastructure
+  tabs via `createSlotExtension`, depending only on the slot owner.
+
+  The slot system in `@checkstack/frontend-api` gains a second type parameter
+  on `createSlot<TContext, TMetadata>` so extensions can declare typed static
+  metadata at registration time (label, icon, access rules, ordering for the
+  infrastructure tab bar). A new `useSlotExtensions(slot)` hook returns typed
+  extensions and subscribes to plugin lifecycle changes.
+
+  Each tab body now stacks a **Runtime** sub-section (live state, read-only)
+  on top of a **Configuration** sub-section (settings, gated by `canUpdate`).
+
+  **Queue runtime panel.** Surfaces aggregated counts (pending / processing /
+  completed / failed) plus three sub-tabs of recent jobs: **Active**, **Recent
+  failed** (with the failure message), and **Recent completed** (with
+  duration). Job payloads are deliberately not surfaced — they may carry
+  secrets and need a separate manage-access gate to be shown.
+
+  To support this, `Queue<T>` gains a required `listJobs(opts)` method
+  returning `JobSummary[]` (no payloads), and `QueueStats` gains a
+  `scope: "instance" | "cluster"` field. The in-memory queue keeps rolling
+  ring buffers (200 entries) for completed/failed history and tracks active
+  jobs by id; BullMQ uses native `getJobs`. `QueueManager.listJobs` aggregates
+  across queues and sorts (most-recent-first for terminal states, FIFO for
+  active/waiting/delayed).
+
+  **Cache runtime panel.** Lists the top N entries by size (or by recency) so
+  operators can debug a cache filling up. Values are deliberately omitted —
+  PII / secret risk. Backends opt in via an optional `listEntries?` method on
+  `CacheProvider`; non-supporting backends return `{ supported: false }` and
+  the UI renders a "not supported by this backend" hint. The in-memory cache
+  implements it using its existing per-entry byte tracking.
+
+  `CacheStats` also gains `scope: "instance" | "cluster"`.
+
+  **Multi-instance scope warning.** A new `<InstanceScopeBanner>` component in
+  `@checkstack/ui` renders a yellow banner above any runtime panel whose
+  backend reports `scope: "instance"` — i.e. in-memory queue or cache running
+  in a horizontally scaled deployment. The banner explains the metrics are
+  local to the responding replica and recommends switching to a clustered
+  backend (Redis-backed queue / cache) for cluster-wide visibility.
+
+  **Bug fix — stable cache provider proxy.** `CacheManagerImpl.getProvider()`
+  now returns a single stable proxy that delegates to whatever provider is
+  currently active. Previously, consumers of `createCachedScope` (and any
+  direct `cacheManager.getProvider()` caller) captured the active provider
+  reference at plugin-init time. After any `setActiveBackend` call — including
+  saving the same memory config in the new Cache tab, which reconstructs the
+  in-memory cache — those scopes wrote to an orphaned old provider while the
+  runtime panel read stats from the new (empty) one, making the runtime panel
+  appear to report 0 keys. With the proxy, all consumers share a single stable
+  identity and writes always land in the active provider.
+
+  **Bytes tracking on the in-memory cache.** `InMemoryCache.getStats().sizeBytes`
+  now returns a running approximation (UTF-8 bytes of the key plus
+  `v8.serialize(value).byteLength`, with a JSON fallback) that's kept in sync
+  across all eviction paths. Treat the number as a sanity gauge; it doesn't
+  include `Map` per-entry overhead.
+
+  **Pagination.** Both `Queue<T>.listJobs` and `CacheProvider.listEntries?`
+  are offset-paginated. Inputs gain an `offset: number`; outputs change to
+  `{ items, total: number | null, hasMore: boolean }`. `total` is nullable
+  so backends that can't compute it cheaply still paginate via `hasMore`.
+  The UI uses the existing `<Pagination>` component with a 25-row default
+  page size. `QueueManager.listJobs` aggregates by over-fetching
+  `[0, offset+limit)` per queue, merge-sorting, then slicing the window —
+  optimal for the single-queue case, acceptable for the multi-queue case
+  within the UI's reasonable page-depth bounds. BullMQ uses native offset
+  ranges via `getJobs(types, start, end)` plus `getJobCounts` for `total`.
+
+  **Pending tab.** The Queue runtime panel exposes a virtual `"pending"`
+  state (waiting ∪ delayed, FIFO). It's now the default sub-tab, since
+  "what's queued up?" is the most common question. Per-row state is shown
+  when viewing the combined list.
+
+  **Recurring schedules visible under Pending.** Cron- and interval-based
+  recurring jobs (e.g. healthchecks) are surfaced under Pending/Delayed
+  between fires, with a `nextRunAt` countdown column and a "(recurring)"
+  label. `JobSummary` gains optional `nextRunAt: Date` and `recurring:
+boolean` fields. The in-memory queue synthesises these rows from its
+  `recurringJobs` registry; BullMQ already materialises the next fire of
+  each scheduler as a delayed job and we now surface its trigger time and
+  the `repeatJobKey`-derived `recurring` flag.
+
+  **Bug fix — drop hook emits with no listeners.** `EventBus.emit` no
+  longer enqueues a job when zero listeners (distributed or instance-local)
+  are registered for the hook. Previously, hooks like
+  `core.plugin.initialized` — emitted on every plugin init but subscribed
+  to by nothing in the core repo — accumulated one waiting job per emit
+  forever. The in-memory queue's `processNext` short-circuits when there
+  are zero consumer groups, so its post-loop cleanup never ran for these
+  orphaned jobs. The fix drops the emit at the source and logs a debug
+  line. Note: in distributed deployments using a Redis-backed queue, this
+  means a subscriber on another replica won't receive an event if no
+  replica that emits it has a local listener. Plugins needing cross-process
+  delivery must register their listener on every replica that should
+  receive the hook.
+
+  **Breaking notes (treated as minor under beta semantics)**:
+
+  - `@checkstack/infrastructure-common` removes `registerInfrastructureTab`
+    and `getInfrastructureTabs`; former callers must register an extension
+    into `InfrastructureTabsSlot`.
+  - `@checkstack/queue-api`'s `Queue<T>` interface requires the new
+    `listJobs(opts)` method returning `ListJobsResult` (paginated). Both
+    bundled queue backends (memory, BullMQ) are updated; out-of-tree
+    implementations will need to add it.
+  - `QueueStats` and `CacheStats` add a required `scope` field.
+  - `CacheProvider.listEntries?` (when implemented) now returns
+    `ListEntriesResult` instead of `CacheEntrySummary[]`.
+  - `JobState` adds a `"pending"` variant.
+
+- Updated dependencies [42abfff]
+- Updated dependencies [aa89bc5]
+  - @checkstack/common@0.9.0
+  - @checkstack/queue-api@0.3.0
+  - @checkstack/cache-api@0.3.0
+  - @checkstack/api-docs-common@0.1.12
+  - @checkstack/auth-common@0.6.6
+  - @checkstack/backend-api@0.15.1
+  - @checkstack/pluginmanager-common@0.2.1
+  - @checkstack/signal-backend@0.2.4
+  - @checkstack/signal-common@0.2.2
+
 ## 0.9.0
 
 ### Minor Changes

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@checkstack/backend",
-  "version": "0.9.0",
+  "version": "0.9.1",
   "license": "Elastic-2.0",
   "checkstack": {
     "type": "backend"
@@ -14,16 +14,16 @@
     "lint:code": "eslint . --max-warnings 0"
   },
   "dependencies": {
-    "@checkstack/api-docs-common": "0.1.10",
-    "@checkstack/auth-common": "0.6.4",
-    "@checkstack/backend-api": "0.14.1",
-    "@checkstack/common": "0.7.0",
-    "@checkstack/drizzle-helper": "0.0.4",
-    "@checkstack/cache-api": "0.2.3",
-    "@checkstack/queue-api": "0.2.17",
-    "@checkstack/signal-backend": "0.2.2",
-    "@checkstack/signal-common": "0.2.0",
-    "@checkstack/pluginmanager-common": "0.1.0",
+    "@checkstack/api-docs-common": "0.1.11",
+    "@checkstack/auth-common": "0.6.5",
+    "@checkstack/backend-api": "0.15.0",
+    "@checkstack/common": "0.8.0",
+    "@checkstack/drizzle-helper": "0.0.5",
+    "@checkstack/cache-api": "0.2.4",
+    "@checkstack/queue-api": "0.2.18",
+    "@checkstack/signal-backend": "0.2.3",
+    "@checkstack/signal-common": "0.2.1",
+    "@checkstack/pluginmanager-common": "0.2.0",
     "@hono/zod-validator": "^0.7.6",
     "@orpc/client": "^1.13.14",
     "@orpc/contract": "^1.13.14",
@@ -44,9 +44,9 @@
     "@types/pg": "^8.11.0",
     "@types/bun": "latest",
     "@types/semver": "^7.5.0",
-    "@checkstack/tsconfig": "0.0.6",
-    "@checkstack/scripts": "0.1.2",
-    "@checkstack/test-utils-backend": "0.1.23",
+    "@checkstack/tsconfig": "0.0.7",
+    "@checkstack/scripts": "0.3.0",
+    "@checkstack/test-utils-backend": "0.1.24",
     "drizzle-kit": "^0.31.10"
   }
 }

package/src/services/cache-manager.test.ts

@@ -0,0 +1,172 @@
+import { describe, it, expect, beforeEach } from "bun:test";
+import { z } from "zod";
+import { CacheManagerImpl } from "./cache-manager";
+import { CachePluginRegistryImpl } from "./cache-plugin-registry";
+import type { CachePlugin, CacheProvider, CacheStats } from "@checkstack/cache-api";
+import type { ConfigService, Logger } from "@checkstack/backend-api";
+import { createMockLogger } from "@checkstack/test-utils-backend";
+
+class FakeInMemoryCache implements CacheProvider {
+  store = new Map<string, unknown>();
+  hits = 0;
+  misses = 0;
+
+  async get<T>(key: string): Promise<T | undefined> {
+    if (this.store.has(key)) {
+      this.hits++;
+      return this.store.get(key) as T;
+    }
+    this.misses++;
+    return undefined;
+  }
+  async set<T>(key: string, value: T): Promise<void> {
+    this.store.set(key, value);
+  }
+  async delete(key: string): Promise<void> {
+    this.store.delete(key);
+  }
+  async deleteByPrefix(): Promise<number> {
+    return 0;
+  }
+  async has(key: string): Promise<boolean> {
+    return this.store.has(key);
+  }
+  async getStats(): Promise<CacheStats> {
+    return {
+      keyCount: this.store.size,
+      sizeBytes: null,
+      hits: this.hits,
+      misses: this.misses,
+      scope: "instance",
+    };
+  }
+}
+
+const memoryConfigSchema = z.record(z.string(), z.unknown());
+
+const createMemoryPlugin = (
+  factory: () => FakeInMemoryCache,
+): CachePlugin<unknown> => ({
+  id: "memory",
+  displayName: "Memory",
+  configVersion: 1,
+  configSchema: memoryConfigSchema,
+  createProvider: () => factory(),
+});
+
+class StubConfigService implements ConfigService {
+  store = new Map<string, unknown>();
+  async set<T>(configId: string, _s: unknown, _v: number, data: T) {
+    this.store.set(configId, data);
+  }
+  async get<T>(configId: string): Promise<T | undefined> {
+    return this.store.get(configId) as T | undefined;
+  }
+  async getRedacted<T>(configId: string): Promise<Partial<T> | undefined> {
+    return this.store.get(configId) as Partial<T> | undefined;
+  }
+  async delete(configId: string): Promise<void> {
+    this.store.delete(configId);
+  }
+  async list(): Promise<string[]> {
+    return [...this.store.keys()];
+  }
+}
+
+describe("CacheManagerImpl provider proxy", () => {
+  let registry: CachePluginRegistryImpl;
+  let configService: StubConfigService;
+  let logger: Logger;
+  let manager: CacheManagerImpl;
+  let instances: FakeInMemoryCache[];
+
+  beforeEach(() => {
+    registry = new CachePluginRegistryImpl();
+    configService = new StubConfigService();
+    logger = createMockLogger() as unknown as Logger;
+    instances = [];
+    registry.register(
+      createMemoryPlugin(() => {
+        const inst = new FakeInMemoryCache();
+        instances.push(inst);
+        return inst;
+      }),
+    );
+    manager = new CacheManagerImpl(registry, configService, logger);
+  });
+
+  it("returns the same proxy reference across calls", async () => {
+    const a = manager.getProvider();
+    const b = manager.getProvider();
+    expect(a).toBe(b);
+  });
+
+  it("delegates writes to the active provider after loadConfiguration", async () => {
+    const provider = manager.getProvider();
+    await manager.loadConfiguration();
+    await provider.set("foo", "bar");
+    expect(instances[0].store.get("foo")).toBe("bar");
+  });
+
+  it("keeps a captured proxy reference live across setActiveBackend", async () => {
+    await manager.loadConfiguration();
+    const captured = manager.getProvider();
+    await captured.set("before", 1);
+
+    await manager.setActiveBackend("memory", {
+      maxEntries: 5000,
+      sweepIntervalMs: 30_000,
+    });
+
+    // The captured reference must now write to the NEW active provider,
+    // not the orphaned old one.
+    await captured.set("after", 2);
+
+    const oldInstance = instances[0];
+    const newInstance = instances.at(-1)!;
+    expect(oldInstance).not.toBe(newInstance);
+    expect(oldInstance.store.has("after")).toBe(false);
+    expect(newInstance.store.has("after")).toBe(true);
+  });
+
+  it("getStats reads from the currently active provider", async () => {
+    await manager.loadConfiguration();
+    const provider = manager.getProvider();
+    await provider.set("k1", "v1");
+    await provider.get("k1"); // hit
+    await provider.get("missing"); // miss
+
+    const stats = await provider.getStats!();
+    expect(stats.keyCount).toBe(1);
+    expect(stats.hits).toBe(1);
+    expect(stats.misses).toBe(1);
+  });
+
+  it("getStats returns all-null when active provider has no getStats impl", async () => {
+    // Plugin whose provider doesn't implement getStats.
+    const minimalPlugin: CachePlugin<unknown> = {
+      id: "noStats",
+      displayName: "No stats",
+      configVersion: 1,
+      configSchema: memoryConfigSchema,
+      createProvider: (): CacheProvider => ({
+        get: async () => undefined,
+        set: async () => {},
+        delete: async () => {},
+        deleteByPrefix: async () => 0,
+        has: async () => false,
+      }),
+    };
+    registry.register(minimalPlugin);
+
+    await manager.setActiveBackend("noStats", {});
+    const stats = await manager.getProvider().getStats!();
+    expect(stats).toEqual({
+      keyCount: null,
+      sizeBytes: null,
+      hits: null,
+      misses: null,
+      scope: "instance",
+    });
+  });
+});

package/src/services/cache-manager.ts

@@ -1,4 +1,10 @@
-import type { CacheManager, CacheProvider } from "@checkstack/cache-api";
+import type {
+  CacheManager,
+  CacheProvider,
+  CacheStats,
+  ListEntriesOptions,
+  ListEntriesResult,
+} from "@checkstack/cache-api";
 import type { CachePluginRegistryImpl } from "./cache-plugin-registry";
 import type { Logger, ConfigService } from "@checkstack/backend-api";
 import { z } from "zod";
@@ -18,6 +24,10 @@ type ActiveCachePointer = z.infer<typeof activeCachePointerSchema>;
 /**
  * A no-op CacheProvider used before any backend is configured.
  * All operations are safe to call but behave as if the cache is empty.
+ *
+ * Held only as the initial value of `activeProvider` until
+ * `loadConfiguration()` runs; consumers always interact with the stable
+ * proxy below, so they pick up the real provider as soon as it's ready.
  */
 const nullProvider: CacheProvider = {
   // eslint-disable-next-line unicorn/no-useless-undefined
@@ -31,8 +41,18 @@ const nullProvider: CacheProvider = {
 /**
  * CacheManagerImpl handles cache provider lifecycle and backend switching.
  *
- * Simpler than QueueManagerImpl — no proxy pattern needed since cache is
- * stateless key/value. The active provider is replaced atomically on backend switch.
+ * `getProvider()` returns a single stable proxy whose methods delegate to
+ * whichever provider is currently active. This means:
+ *
+ *  - Plugins that capture the proxy at init time (e.g. via
+ *    `createCachedScope`) keep functioning across backend switches and
+ *    don't write into orphaned providers.
+ *  - The Infrastructure runtime panel reads stats from the same instance
+ *    plugins write to — no split-brain.
+ *
+ * The earlier "no proxy pattern needed since cache is stateless key/value"
+ * comment was wrong: in-memory backends are stateful and replacing the
+ * active reference would orphan their state.
  */
 export class CacheManagerImpl implements CacheManager {
   private activePluginId: string = "memory";
@@ -42,12 +62,49 @@ export class CacheManagerImpl implements CacheManager {
   };
   private configVersion: number = 0;
   private activeProvider: CacheProvider = nullProvider;
+  private readonly providerProxy: CacheProvider;
 
   constructor(
     private registry: CachePluginRegistryImpl,
     private configService: ConfigService,
     private logger: Logger,
-  ) {}
+  ) {
+    this.providerProxy = this.createProviderProxy();
+  }
+
+  private createProviderProxy(): CacheProvider {
+    return {
+      get: <T>(key: string) => this.activeProvider.get<T>(key),
+      set: <T>(key: string, value: T, ttlMs?: number) =>
+        this.activeProvider.set<T>(key, value, ttlMs),
+      delete: (key: string) => this.activeProvider.delete(key),
+      deleteByPrefix: (prefix: string) =>
+        this.activeProvider.deleteByPrefix(prefix),
+      has: (key: string) => this.activeProvider.has(key),
+      getStats: async (): Promise<CacheStats> => {
+        const provider = this.activeProvider;
+        if (provider.getStats) {
+          return provider.getStats();
+        }
+        return {
+          keyCount: null,
+          sizeBytes: null,
+          hits: null,
+          misses: null,
+          scope: "instance",
+        };
+      },
+      listEntries: async (
+        opts: ListEntriesOptions,
+      ): Promise<ListEntriesResult> => {
+        const provider = this.activeProvider;
+        if (!provider.listEntries) {
+          return { items: [], total: 0, hasMore: false };
+        }
+        return provider.listEntries(opts);
+      },
+    };
+  }
 
   async loadConfiguration(): Promise<void> {
     try {
@@ -83,11 +140,10 @@ export class CacheManagerImpl implements CacheManager {
         );
       }
 
-      // Initialize the active provider
       this.initializeProvider();
     } catch (error) {
       this.logger.error("Failed to load cache configuration", error);
-      // Continue with defaults — nullProvider is already set
+      // Continue with defaults — nullProvider is already set behind the proxy.
     }
   }
 
@@ -113,8 +169,12 @@ export class CacheManagerImpl implements CacheManager {
     }
   }
 
+  /**
+   * Returns the stable proxy. The same instance is returned for the
+   * lifetime of the manager, so callers can safely capture it once.
+   */
   getProvider(): CacheProvider {
-    return this.activeProvider;
+    return this.providerProxy;
   }
 
   getActivePlugin(): string {
@@ -126,21 +186,17 @@ export class CacheManagerImpl implements CacheManager {
   }
 
   async setActiveBackend(pluginId: string, config: unknown): Promise<void> {
-    // 1. Validate plugin exists
     const newPlugin = this.registry.getPlugin(pluginId);
     if (!newPlugin) {
       throw new Error(`Cache plugin '${pluginId}' not found`);
     }
 
-    // 2. Validate config against schema
     newPlugin.configSchema.parse(config);
 
-    // 3. Create new provider (acts as connection test)
     this.logger.info("🔍 Testing new cache provider...");
     let newProvider: CacheProvider;
     try {
       newProvider = newPlugin.createProvider(config, this.logger);
-      // Quick smoke test
       await newProvider.set("__test__", true, 1000);
       await newProvider.delete("__test__");
       this.logger.info("✅ Cache provider test successful");
@@ -150,20 +206,17 @@ export class CacheManagerImpl implements CacheManager {
       throw new Error(`Failed to create cache provider: ${message}`);
     }
 
-    // 4. Stop old provider if it has a stop method
     const oldProvider = this.activeProvider;
     if ("stop" in oldProvider && typeof oldProvider.stop === "function") {
       await (oldProvider as { stop: () => Promise<void> }).stop();
     }
 
-    // 5. Switch to new provider
     const oldPluginId = this.activePluginId;
     this.activePluginId = pluginId;
     this.activeConfig = config;
     this.configVersion++;
     this.activeProvider = newProvider;
 
-    // 6. Persist configuration
     await this.configService.set(
       pluginId,
       newPlugin.configSchema,

package/src/services/event-bus.test.ts

@@ -465,5 +465,57 @@ describe("EventBus", () => {
         `No local listeners for hook: ${testHook.id}`
       );
     });
+
+    it("should drop emit (not enqueue) when no listeners are registered", async () => {
+      const testHook = createHook<{ value: number }>(
+        "test.emit.no.listeners",
+      );
+
+      // Capture queue creation: getQueue is invoked lazily on enqueue.
+      const before = (mockQueueManager as any).getQueue?.mock?.calls?.length ?? 0;
+
+      await eventBus.emit(testHook, { value: 1 });
+      await eventBus.emit(testHook, { value: 2 });
+
+      const after = (mockQueueManager as any).getQueue?.mock?.calls?.length ?? 0;
+      // No queue should be created and no enqueue should occur for an
+      // entirely unsubscribed hook — otherwise jobs would pile up forever.
+      expect(after).toBe(before);
+
+      expect(mockLogger.debug).toHaveBeenCalledWith(
+        `Dropped hook ${testHook.id}: no listeners registered`,
+      );
+    });
+
+    it("should enqueue when at least one distributed listener is registered", async () => {
+      const testHook = createHook<{ value: number }>(
+        "test.emit.with.listener",
+      );
+      const received: number[] = [];
+      await eventBus.subscribe("test-plugin", testHook, async (payload) => {
+        received.push(payload.value);
+      });
+
+      await eventBus.emit(testHook, { value: 7 });
+      // Allow the mock queue to deliver synchronously.
+      await new Promise((resolve) => setTimeout(resolve, 10));
+      expect(received).toContain(7);
+    });
+
+    it("should enqueue when at least one instance-local listener is registered", async () => {
+      const testHook = createHook<{ value: number }>(
+        "test.emit.with.local.listener",
+      );
+      await eventBus.subscribe("test-plugin", testHook, async () => {}, {
+        mode: "instance-local",
+      });
+
+      // emit (distributed) should still enqueue because a local listener
+      // exists — drops are based on absence of *any* listener.
+      await eventBus.emit(testHook, { value: 1 });
+      expect(mockLogger.debug).not.toHaveBeenCalledWith(
+        `Dropped hook ${testHook.id}: no listeners registered`,
+      );
+    });
   });
 });

package/src/services/event-bus.ts

@@ -185,9 +185,35 @@ export class EventBus implements IEventBus {
   }
 
   /**
-   * Emit a hook
+   * Emit a hook.
+   *
+   * Skips the underlying queue enqueue when no listener has been
+   * registered for the hook in this process. Without this guard, hooks
+   * with no subscribers (e.g. `core.plugin.initialized` when no plugin
+   * has registered a listener) would accumulate jobs in the in-memory
+   * queue forever — they'd be enqueued, no consumer group exists to
+   * process them, and `processNext` short-circuits before its cleanup
+   * pass when `consumerGroups.size === 0`.
+   *
+   * Note: this checks listeners in the *local process*. In distributed
+   * deployments with a Redis-backed queue, a subscriber on another
+   * replica would never see the event under this rule. Callers that
+   * need cross-process delivery must therefore ensure at least one
+   * listener registers on every replica that should receive the hook.
    */
   async emit<T>(hook: Hook<T>, payload: T): Promise<void> {
+    const hasDistributedListeners =
+      (this.listeners.get(hook.id)?.length ?? 0) > 0;
+    const hasLocalListeners =
+      (this.localListeners.get(hook.id)?.length ?? 0) > 0;
+
+    if (!hasDistributedListeners && !hasLocalListeners) {
+      this.logger.debug(
+        `Dropped hook ${hook.id}: no listeners registered`,
+      );
+      return;
+    }
+
     let channel = this.queueChannels.get(hook.id);
 
     // Create channel lazily if not exists

package/src/services/queue-manager.ts

@@ -4,6 +4,9 @@ import type {
   SwitchResult,
   RecurringJobInfo,
   QueueStats,
+  JobSummary,
+  ListJobsOptions,
+  ListJobsResult,
 } from "@checkstack/queue-api";
 import type { QueuePluginRegistryImpl } from "./queue-plugin-registry";
 import type { Logger, ConfigService } from "@checkstack/backend-api";
@@ -276,7 +279,11 @@ export class QueueManagerImpl implements QueueManager {
   }
 
   async getAggregatedStats(): Promise<QueueStats> {
-    const aggregated: QueueStats = {
+    // The narrowest scope wins: if any queue reports `instance`, the
+    // aggregate cannot claim cluster-wide accuracy.
+    let scope: "instance" | "cluster" = "cluster";
+    let sawAny = false;
+    const aggregated: Omit<QueueStats, "scope"> = {
       pending: 0,
       processing: 0,
       completed: 0,
@@ -294,13 +301,81 @@ export class QueueManagerImpl implements QueueManager {
           aggregated.completed += stats.completed;
           aggregated.failed += stats.failed;
           aggregated.consumerGroups += stats.consumerGroups;
+          if (stats.scope === "instance") {
+            scope = "instance";
+          }
+          sawAny = true;
+        }
+      } catch {
+        // Queue may not be initialized yet
+      }
+    }
+
+    return { ...aggregated, scope: sawAny ? scope : "instance" };
+  }
+
+  /**
+   * Aggregate `listJobs` across every queue proxy. For deployments with a
+   * single queue (the common case) this is a straight pass-through; with
+   * multiple queues we over-fetch [0, offset+limit) per queue, merge-sort
+   * by the same key the per-queue impls use, then slice the requested
+   * window. Sort key:
+   *  - completed/failed → finishedAt desc (most-recent first)
+   *  - active           → startedAt asc (oldest still-running first)
+   *  - waiting/delayed  → enqueuedAt asc (FIFO)
+   */
+  async listJobs(opts: ListJobsOptions): Promise<ListJobsResult> {
+    const limit = Math.max(0, opts.limit);
+    const offset = Math.max(0, opts.offset);
+    if (limit === 0) {
+      return { items: [], total: 0, hasMore: false };
+    }
+
+    const fetchUpTo = offset + limit;
+    const merged: JobSummary[] = [];
+    let totalSum = 0;
+    let totalIsNull = false;
+
+    for (const proxy of this.queueProxies.values()) {
+      try {
+        const delegate = proxy.getDelegate();
+        if (delegate) {
+          const part = await delegate.listJobs({
+            state: opts.state,
+            offset: 0,
+            limit: fetchUpTo,
+          });
+          merged.push(...part.items);
+          if (part.total === null) {
+            totalIsNull = true;
+          } else {
+            totalSum += part.total;
+          }
         }
       } catch {
         // Queue may not be initialized yet
       }
     }
 
-    return aggregated;
+    const sortKey = (j: JobSummary): number => {
+      if (opts.state === "completed" || opts.state === "failed") {
+        return -(j.finishedAt?.getTime() ?? 0);
+      }
+      if (opts.state === "active") {
+        return j.startedAt?.getTime() ?? 0;
+      }
+      return j.enqueuedAt.getTime();
+    };
+    merged.sort((a, b) => sortKey(a) - sortKey(b));
+
+    const items = merged.slice(offset, offset + limit);
+    const total = totalIsNull ? null : totalSum;
+    const hasMore =
+      total === null
+        ? merged.length > offset + items.length
+        : offset + items.length < total;
+
+    return { items, total, hasMore };
   }
 
   async listAllRecurringJobs(): Promise<RecurringJobInfo[]> {

package/src/services/queue-proxy.ts

@@ -5,6 +5,8 @@ import type {
   QueueStats,
   RecurringJobDetails,
   RecurringSchedule,
+  ListJobsOptions,
+  ListJobsResult,
 } from "@checkstack/queue-api";
 import { rootLogger } from "../logger";
 
@@ -173,6 +175,11 @@ export class QueueProxy<T = unknown> implements Queue<T> {
     return delegate.getStats();
   }
 
+  async listJobs(opts: ListJobsOptions): Promise<ListJobsResult> {
+    const delegate = this.ensureDelegate();
+    return delegate.listJobs(opts);
+  }
+
   /**
    * Get subscription count (for testing/debugging)
    */