[RHIDP-13061] Add Rate Limiting for Lightspeed Plugin (#3531)

* feat: add rate limiting for lightspeed and notebooks

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* update lightspeed-backend readme with rate limiting info

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* update changeset to only backend

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* extend express request type with rate limit

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* handle negative and float values for rate limit

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* refactor(notebooks): move permission check to route level to preserve rate limit -> permission check ordering to reduce redundant permission checks

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* migrate from lightspeed.* to intelligent-assistant-*

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

* add example block to lightspeed-backend app-config for rate limiting

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

---------

Signed-off-by: Jordan Dubrick <jdubrick@redhat.com>

Author: Jdubrick

workspaces/lightspeed/.changeset/five-pots-change.md

@@ -0,0 +1,5 @@
+---
+'@red-hat-developer-hub/backstage-plugin-lightspeed-backend': minor
+---
+
+add rate limiting to lightspeed and notebooks

workspaces/lightspeed/plugins/lightspeed-backend/README.md

@@ -65,6 +65,43 @@ intelligent-assistant:
   mcpServers: # Optional - one or more MCP servers
     - name: <mcp server name> # must match the name configured in LCS
       token: ${MCP_TOKEN}
+  rateLimit: # Optional - per-user request rate limits (defaults apply if omitted)
+    expensive:
+      max: 25 # Max requests per minute per user for expensive endpoints (default: 25). Set to 0 to disable.
+    general:
+      max: 200 # Max requests per minute per user for other authenticated endpoints (default: 200). Set to 0 to disable.
+```
+
+#### Rate limiting
+
+The backend applies per-user rate limits to authenticated endpoints as an abuse
+prevention measure. Limits are keyed by the authenticated user's entity ref and
+use a fixed 1-minute window.
+
+**Tiers**:
+
+- **Expensive** (default: 25 requests/minute per user): `POST /v1/query`, and
+  (when Notebooks is enabled) notebook document uploads and RAG queries.
+- **General** (default: 200 requests/minute per user): all other authenticated
+  endpoints, including conversation listing, MCP server management, feedback,
+  and notebook session CRUD.
+- **Excluded**: `/health` and `/notebooks/health` are not rate limited.
+
+When a limit is exceeded, the API returns `429 Too Many Requests` with a
+`Retry-After` header and a JSON error body (`RateLimitExceeded`).
+
+Set `max: 0` on a tier to disable rate limiting for that tier. If the entire
+`rateLimit` block is omitted, the defaults above apply.
+
+**Example** — tighter limits for a small deployment:
+
+```yaml
+intelligent-assistant:
+  rateLimit:
+    expensive:
+      max: 10
+    general:
+      max: 100
 ```
 
 #### MCP servers settings endpoints

workspaces/lightspeed/plugins/lightspeed-backend/app-config.yaml

@@ -1,18 +1,23 @@
 # OPTIONAL: Backend-only configurations
-#intelligent-assistant:
+# intelligent-assistant:
 #  servicePort: 8080 # OPTIONAL: Port for lightspeed-core service (default: 8080)
 #  systemPrompt: <custom_system_prompt> # OPTIONAL: Override default RHDH system prompt
-#
+#  # Optional: Per-user request rate limits (defaults apply if omitted)
+#  rateLimit:
+#   expensive:
+#     max: 10
+#   general:
+#     max: 100
 #  # AI Notebooks (Developer Preview) - Disabled by default
 #  notebooks:
 #    enabled: false # Set to true to enable AI Notebooks feature
-#
+
 #    # REQUIRED when enabled: Query defaults for RAG queries
 #    # Both model and provider_id must be configured together
 #    queryDefaults:
 #      provider_id: ollama # AI provider for query model (e.g., ollama, vllm)
 #      model: llama3.1-8b-instruct # Model to use for answering queries
-#
+
 #    # OPTIONAL: Chunking strategy for document processing
 #    chunkingStrategy:
 #      type: auto # 'auto' (default) or 'static'

workspaces/lightspeed/plugins/lightspeed-backend/config.d.ts

@@ -49,6 +49,38 @@ export interface Config {
        */
       token?: string;
     }>;
+    /**
+     * Per-user rate limiting for Lightspeed API endpoints.
+     * @visibility backend
+     */
+    rateLimit?: {
+      /**
+       * Limits for expensive endpoints (LLM queries, document uploads).
+       * @visibility backend
+       */
+      expensive?: {
+        /**
+         * Maximum requests per minute per user.
+         * Set to 0 to disable rate limiting for this tier.
+         * @default 25
+         * @visibility backend
+         */
+        max?: number;
+      };
+      /**
+       * Limits for all other authenticated endpoints.
+       * @visibility backend
+       */
+      general?: {
+        /**
+         * Maximum requests per minute per user.
+         * Set to 0 to disable rate limiting for this tier.
+         * @default 200
+         * @visibility backend
+         */
+        max?: number;
+      };
+    };
     /**
      * Configuration for AI Notebooks (Developer Preview)
      */

workspaces/lightspeed/plugins/lightspeed-backend/package.json

@@ -52,6 +52,7 @@
     "@langchain/openai": "^0.6.0",
     "@red-hat-developer-hub/backstage-plugin-lightspeed-common": "workspace:^",
     "express": "^4.21.1",
+    "express-rate-limit": "^8.2.2",
     "form-data": "^4.0.5",
     "htmlparser2": "^9.1.0",
     "http-proxy-middleware": "^3.0.2",

workspaces/lightspeed/plugins/lightspeed-backend/src/service/constant.ts

@@ -25,6 +25,13 @@ export const DEFAULT_LIGHTSPEED_SERVICE_HOST = '127.0.0.1'; // Lightspeed core s
 export const DEFAULT_LIGHTSPEED_SERVICE_PORT = 8080; // Lightspeed service port
 export const DEFAULT_MAX_FILE_SIZE_MB = 20 * 1024 * 1024; // 20MB
 
+/**
+ * Rate limiting defaults (window is fixed at 1 minute)
+ */
+export const RATE_LIMIT_WINDOW_MS = 60000;
+export const DEFAULT_EXPENSIVE_RATE_LIMIT_MAX = 25;
+export const DEFAULT_GENERAL_RATE_LIMIT_MAX = 200;
+
 /**
  * Input validation limits for query endpoints
  */

workspaces/lightspeed/plugins/lightspeed-backend/src/service/express.d.ts

@@ -16,10 +16,13 @@
 
 import type { BackstageCredentials } from '@backstage/backend-plugin-api';
 
+import type { RateLimitInfo } from 'express-rate-limit';
+
 // Populated by the identity middleware for use in route handlers.
 declare module 'express-serve-static-core' {
   interface Request {
     credentials?: BackstageCredentials;
     userEntityRef?: string;
+    rateLimit?: RateLimitInfo;
   }
 }

workspaces/lightspeed/plugins/lightspeed-backend/src/service/middleware/createRateLimitMiddleware.test.ts

@@ -0,0 +1,195 @@
+/*
+ * Copyright Red Hat, Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import { mockServices } from '@backstage/backend-test-utils';
+
+import express from 'express';
+import request from 'supertest';
+
+import {
+  DEFAULT_EXPENSIVE_RATE_LIMIT_MAX,
+  DEFAULT_GENERAL_RATE_LIMIT_MAX,
+} from '../constant';
+import {
+  createRateLimitMiddleware,
+  getRateLimitMax,
+} from './createRateLimitMiddleware';
+
+describe('getRateLimitMax', () => {
+  it('returns default expensive max when config is omitted', () => {
+    const config = mockServices.rootConfig({ data: {} });
+    expect(getRateLimitMax(config, 'expensive')).toBe(
+      DEFAULT_EXPENSIVE_RATE_LIMIT_MAX,
+    );
+  });
+
+  it('returns default general max when config is omitted', () => {
+    const config = mockServices.rootConfig({ data: {} });
+    expect(getRateLimitMax(config, 'general')).toBe(
+      DEFAULT_GENERAL_RATE_LIMIT_MAX,
+    );
+  });
+
+  it('returns configured max values when provided', () => {
+    const config = mockServices.rootConfig({
+      data: {
+        'intelligent-assistant': {
+          rateLimit: {
+            expensive: { max: 10 },
+            general: { max: 50 },
+          },
+        },
+      },
+    });
+
+    expect(getRateLimitMax(config, 'expensive')).toBe(10);
+    expect(getRateLimitMax(config, 'general')).toBe(50);
+  });
+
+  it('treats negative values as disabled (0)', () => {
+    const config = mockServices.rootConfig({
+      data: {
+        'intelligent-assistant': {
+          rateLimit: {
+            expensive: { max: -1 },
+            general: { max: -5 },
+          },
+        },
+      },
+    });
+
+    expect(getRateLimitMax(config, 'expensive')).toBe(0);
+    expect(getRateLimitMax(config, 'general')).toBe(0);
+  });
+
+  it('floors decimal values to integers', () => {
+    const config = mockServices.rootConfig({
+      data: {
+        'intelligent-assistant': {
+          rateLimit: {
+            expensive: { max: 10.7 },
+            general: { max: 50.3 },
+          },
+        },
+      },
+    });
+
+    expect(getRateLimitMax(config, 'expensive')).toBe(10);
+    expect(getRateLimitMax(config, 'general')).toBe(50);
+  });
+});
+
+describe('createRateLimitMiddleware', () => {
+  function createTestApp(
+    max: number,
+    tier: 'expensive' | 'general' = 'general',
+  ) {
+    const app = express();
+    const config = mockServices.rootConfig({
+      data: {
+        'intelligent-assistant': {
+          rateLimit: {
+            [tier]: { max },
+          },
+        },
+      },
+    });
+
+    app.use((req, _res, next) => {
+      req.credentials = { $$type: '@backstage/BackstageCredentials' } as any;
+      req.userEntityRef = 'user:default/test-user';
+      next();
+    });
+    app.get('/test', createRateLimitMiddleware(config, tier), (_req, res) => {
+      res.json({ ok: true });
+    });
+
+    return app;
+  }
+
+  it('allows requests up to the configured max', async () => {
+    const app = createTestApp(1);
+
+    const first = await request(app).get('/test');
+    expect(first.status).toBe(200);
+    expect(first.body).toEqual({ ok: true });
+  });
+
+  it('returns 429 with Retry-After when limit is exceeded', async () => {
+    const app = createTestApp(1);
+
+    await request(app).get('/test');
+    const second = await request(app).get('/test');
+
+    expect(second.status).toBe(429);
+    expect(second.headers['retry-after']).toBeDefined();
+    expect(second.body).toEqual({
+      error: {
+        name: 'RateLimitExceeded',
+        message: 'Too many requests. Please try again later.',
+        retryAfter: expect.any(Number),
+      },
+    });
+  });
+
+  it('does not rate limit when max is 0', async () => {
+    const app = createTestApp(0);
+
+    const first = await request(app).get('/test');
+    const second = await request(app).get('/test');
+
+    expect(first.status).toBe(200);
+    expect(second.status).toBe(200);
+  });
+
+  it('tracks limits independently per user', async () => {
+    const config = mockServices.rootConfig({
+      data: {
+        'intelligent-assistant': {
+          rateLimit: {
+            general: { max: 1 },
+          },
+        },
+      },
+    });
+    const app = express();
+
+    app.use((req, _res, next) => {
+      req.credentials = { $$type: '@backstage/BackstageCredentials' } as any;
+      req.userEntityRef =
+        req.headers['x-test-user']?.toString() ?? 'user:default/user-a';
+      next();
+    });
+    app.get(
+      '/test',
+      createRateLimitMiddleware(config, 'general'),
+      (_req, res) => {
+        res.json({ ok: true });
+      },
+    );
+
+    await request(app).get('/test').set('x-test-user', 'user:default/user-a');
+    const blockedForA = await request(app)
+      .get('/test')
+      .set('x-test-user', 'user:default/user-a');
+    const allowedForB = await request(app)
+      .get('/test')
+      .set('x-test-user', 'user:default/user-b');
+
+    expect(blockedForA.status).toBe(429);
+    expect(allowedForB.status).toBe(200);
+  });
+});