diff --git a/.github/workflows/test-evals-instance-ai.yml b/.github/workflows/test-evals-instance-ai.yml
index bde299ea0e1..9356b26013f 100644
--- a/.github/workflows/test-evals-instance-ai.yml
+++ b/.github/workflows/test-evals-instance-ai.yml
@@ -74,7 +74,7 @@ jobs:
 
       - name: Start sandbox service
         if: ${{ inputs.sandbox-provider == 'n8n-sandbox' }}
-        run: pnpm tsx packages/testing/containers/start-sandbox.ts --network n8n-eval-net
+        run: pnpm --filter n8n-containers services --services sandbox --network n8n-eval-net --name n8n-svc-sandbox
 
       - name: Start n8n containers
         env:
@@ -284,14 +284,13 @@ jobs:
           done
 
           # Sandbox service container logs (when using n8n-sandbox provider)
-          for c in sandbox-api sandbox-runner-1; do
-            if docker ps -aq --filter "name=$c" | grep -q .; then
-              echo ""
-              echo "============================================================"
-              echo "=== $c (last 100 lines) ==="
-              echo "============================================================"
-              docker logs "$c" 2>&1 | tail -100 || true
-            fi
+          for c in $(docker ps -aq --filter "label=com.docker.compose.project=n8n-svc-sandbox"); do
+            name=$(docker inspect --format '{{.Name}}' "$c" | sed 's|^/||')
+            echo ""
+            echo "============================================================"
+            echo "=== $name (last 100 lines) ==="
+            echo "============================================================"
+            docker logs "$c" 2>&1 | tail -100 || true
           done
 
       - name: Stop n8n containers
@@ -302,9 +301,8 @@ jobs:
             docker stop "${ids[@]}" 2>/dev/null || true
             docker rm "${ids[@]}" 2>/dev/null || true
           fi
-          # Sandbox service cleanup (safe even if containers don't exist)
-          docker stop sandbox-runner-1 sandbox-api 2>/dev/null || true
-          docker rm sandbox-runner-1 sandbox-api 2>/dev/null || true
+          # Sandbox service cleanup
+          pnpm --filter n8n-containers services:clean 2>/dev/null || true
           docker network rm n8n-eval-net 2>/dev/null || true
 
       - name: Post eval results to PR
diff --git a/packages/testing/containers/n8n-start-stack.ts b/packages/testing/containers/n8n-start-stack.ts
index c2d994c6f32..de66096b21c 100644
--- a/packages/testing/containers/n8n-start-stack.ts
+++ b/packages/testing/containers/n8n-start-stack.ts
@@ -63,6 +63,7 @@ ${colors.yellow}Options:${colors.reset}
   --mains <n>       Number of main instances (default: 1)
   --workers <n>     Number of worker instances (default: 1)
   --name <name>     Project name for parallel runs
+  --network <name>  Docker network name (default: auto-generated)
   --env KEY=VALUE   Set environment variables
   --plan <plan>     Use performance plan preset (${Object.keys(BASE_PERFORMANCE_PLANS).join(', ')})
   --help, -h        Show this help
@@ -154,6 +155,7 @@ async function main() {
 			mains: { type: 'string' },
 			workers: { type: 'string' },
 			name: { type: 'string' },
+			network: { type: 'string' },
 			env: { type: 'string', multiple: true },
 			plan: { type: 'string' },
 		},
@@ -199,6 +201,7 @@ async function main() {
 			(servicesOnly
 				? `n8n-svc-${Math.random().toString(36).substring(7)}`
 				: `n8n-stack-${Math.random().toString(36).substring(7)}`),
+		networkName: values.network,
 	};
 
 	// Handle queue mode (mains > 1 or workers > 0)
@@ -273,6 +276,7 @@ async function main() {
 			const stack = await createServiceStack({
 				services,
 				projectName: config.projectName,
+				networkName: config.networkName,
 			});
 
 			const envVars = writeDevEnvFile(stack, services);
diff --git a/packages/testing/containers/service-stack.ts b/packages/testing/containers/service-stack.ts
index 7879629aa2c..196a6cab886 100644
--- a/packages/testing/containers/service-stack.ts
+++ b/packages/testing/containers/service-stack.ts
@@ -8,6 +8,7 @@ import { createN8NStack, type N8NStack } from './stack';
 export interface ServiceStackOptions {
 	services: ServiceName[];
 	projectName?: string;
+	networkName?: string;
 }
 
 /**
@@ -22,7 +23,7 @@ export interface ServiceStackOptions {
  * await stack.stop();
  */
 export async function createServiceStack(options: ServiceStackOptions): Promise<N8NStack> {
-	const { services, projectName } = options;
+	const { services, projectName, networkName } = options;
 
 	return await createN8NStack({
 		mains: 0,
@@ -30,6 +31,7 @@ export async function createServiceStack(options: ServiceStackOptions): Promise<
 		postgres: services.includes('postgres'),
 		services,
 		projectName,
+		networkName,
 		external: true,
 	});
 }
diff --git a/packages/testing/containers/services/types.ts b/packages/testing/containers/services/types.ts
index fbde9b38a20..c5864b8369d 100644
--- a/packages/testing/containers/services/types.ts
+++ b/packages/testing/containers/services/types.ts
@@ -77,6 +77,8 @@ export interface StackConfig {
 	services?: readonly ServiceName[];
 	/** When true, services target host machine instead of Docker-internal n8n */
 	external?: boolean;
+	/** When set, the Docker network uses this exact name instead of a random UUID. */
+	networkName?: string;
 	/**
 	 * Caddy load-balancer upstream-selection policy. Only applies when `mains > 1`.
 	 * Defaults to `'first'` — sticky to main #1, useful for UI debuggability.
diff --git a/packages/testing/containers/stack.ts b/packages/testing/containers/stack.ts
index c269e246827..f08cf0fc89e 100644
--- a/packages/testing/containers/stack.ts
+++ b/packages/testing/containers/stack.ts
@@ -86,6 +86,7 @@ export async function createN8NStack(config: N8NConfig = {}): Promise<N8NStack>
 		workerResourceQuota,
 		services: enabledServices = [],
 		external = false,
+		networkName,
 	} = config;
 
 	const log = createElapsedLogger('stack');
@@ -115,7 +116,8 @@ export async function createN8NStack(config: N8NConfig = {}): Promise<N8NStack>
 	let network: StartedNetwork;
 	try {
 		const networkStart = performance.now();
-		network = await new Network().start();
+		const uuid = networkName ? { nextUuid: () => networkName } : undefined;
+		network = await new Network(uuid).start();
 		telemetry.recordNetwork(Math.round(performance.now() - networkStart));
 	} catch (error) {
 		const message = error instanceof Error ? error.message : String(error);
diff --git a/packages/testing/containers/start-sandbox.ts b/packages/testing/containers/start-sandbox.ts
deleted file mode 100644
index e63eeeec965..00000000000
--- a/packages/testing/containers/start-sandbox.ts
+++ /dev/null
@@ -1,45 +0,0 @@
-#!/usr/bin/env tsx
-/**
- * Standalone script to start the n8n sandbox service (API + runner) using the
- * testcontainers-based sandbox service definition. Creates a named Docker
- * network so that separately-started n8n containers can join it.
- *
- * Usage: pnpm tsx packages/testing/containers/start-sandbox.ts [--network <name>]
- */
-import { parseArgs } from 'node:util';
-import { Network } from 'testcontainers';
-import type { Uuid } from 'testcontainers';
-
-import { sandbox } from './services/sandbox';
-
-const { values } = parseArgs({
-	options: { network: { type: 'string', default: 'n8n-eval-net' } },
-	strict: false,
-});
-
-const networkName = values.network ?? 'n8n-eval-net';
-const projectName = 'n8n-sandbox-ci';
-
-class FixedName implements Uuid {
-	constructor(private readonly name: string) {}
-	nextUuid() {
-		return this.name;
-	}
-}
-
-async function main() {
-	console.log(`Creating network "${networkName}"...`);
-	const network = await new Network(new FixedName(networkName)).start();
-
-	console.log('Starting sandbox service...');
-	const result = await sandbox.start(network, projectName);
-
-	console.log(`Sandbox API URL: ${result.meta.apiUrl}`);
-	console.log(`Network: ${network.getName()}`);
-	console.log('Sandbox service is ready');
-}
-
-main().catch((error) => {
-	console.error(error);
-	process.exit(1);
-});