diff --git a/internal-packages/benchmark-utils/package.json b/internal-packages/benchmark-utils/package.json
index d2b8e62..7922767 100644
--- a/internal-packages/benchmark-utils/package.json
+++ b/internal-packages/benchmark-utils/package.json
@@ -4,6 +4,7 @@
   "private": true,
   "type": "module",
   "exports": {
+    "./run-build": "./src/run-build.ts",
     "./seeded-random": "./src/seeded-random.ts"
   },
   "devDependencies": {
diff --git a/internal-packages/benchmark-utils/src/run-build.ts b/internal-packages/benchmark-utils/src/run-build.ts
new file mode 100644
index 0000000..feb8381
--- /dev/null
+++ b/internal-packages/benchmark-utils/src/run-build.ts
@@ -0,0 +1,33 @@
+import { execSync } from 'node:child_process'
+import { Buffer } from 'node:buffer'
+import { delimiter, resolve } from 'node:path'
+
+function asString(value: unknown): string {
+  if (typeof value === 'string') return value
+  if (Buffer.isBuffer(value)) return value.toString('utf8')
+  return ''
+}
+
+export function runBuild(configName: string, baseDir: string): void {
+  const localBin = resolve(baseDir, 'node_modules/.bin')
+  const env = {
+    ...process.env,
+    PATH: `${localBin}${delimiter}${process.env.PATH ?? ''}`,
+  }
+  try {
+    execSync(`rolldown -c configs/${configName}.ts`, {
+      cwd: baseDir,
+      stdio: 'pipe',
+      env,
+    })
+  } catch (err) {
+    const isObject = typeof err === 'object' && err !== null
+    const status = isObject && 'status' in err && typeof err.status === 'number' ? err.status : null
+    const stdout = asString(isObject && 'stdout' in err ? err.stdout : undefined)
+    const stderr = asString(isObject && 'stderr' in err ? err.stderr : undefined)
+    throw new Error(
+      `build:${configName} failed (exit ${status ?? 'unknown'}):\n${stdout}${stderr}`,
+      { cause: err },
+    )
+  }
+}
diff --git a/packages/emotion/benchmark/bench/emotion.bench.ts b/packages/emotion/benchmark/bench/emotion.bench.ts
index 373d192..46a279f 100644
--- a/packages/emotion/benchmark/bench/emotion.bench.ts
+++ b/packages/emotion/benchmark/bench/emotion.bench.ts
@@ -1,13 +1,18 @@
 import { bench, describe } from 'vitest'
 import { execSync } from 'node:child_process'
-import { existsSync, rmSync } from 'node:fs'
+import { existsSync, readdirSync, rmSync } from 'node:fs'
 import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
 
 const baseDir = resolve(import.meta.dirname, '..')
 const distBase = resolve(baseDir, 'dist')
 const componentsDir = resolve(baseDir, 'shared-app/src/components')
+const expectedComponents = 100
 
-if (!existsSync(componentsDir)) {
+const currentComponents = existsSync(componentsDir)
+  ? readdirSync(componentsDir).filter((f) => f.endsWith('.tsx')).length
+  : 0
+if (currentComponents !== expectedComponents) {
   execSync('pnpm generate', { cwd: baseDir, stdio: 'inherit' })
 }
 
@@ -18,18 +23,11 @@ function cleanDist(name: string) {
   }
 }
 
-function runBuild(name: string) {
-  execSync(`rolldown -c configs/${name}.ts`, {
-    cwd: baseDir,
-    stdio: 'pipe',
-  })
-}
-
 describe('Emotion Benchmark', () => {
   bench(
     '@rolldown/plugin-emotion',
     () => {
-      runBuild('custom')
+      runBuild('custom', baseDir)
     },
     { teardown: () => cleanDist('custom') },
   )
@@ -37,7 +35,7 @@ describe('Emotion Benchmark', () => {
   bench(
     '@rolldown/plugin-babel',
     () => {
-      runBuild('babel')
+      runBuild('babel', baseDir)
     },
     { teardown: () => cleanDist('babel') },
   )
@@ -45,7 +43,7 @@ describe('Emotion Benchmark', () => {
   bench(
     '@rollup/plugin-swc',
     () => {
-      runBuild('swc')
+      runBuild('swc', baseDir)
     },
     { teardown: () => cleanDist('swc') },
   )
diff --git a/packages/emotion/benchmark/bench/emotion.test.ts b/packages/emotion/benchmark/bench/emotion.test.ts
new file mode 100644
index 0000000..cbb6865
--- /dev/null
+++ b/packages/emotion/benchmark/bench/emotion.test.ts
@@ -0,0 +1,21 @@
+import { describe, test } from 'vitest'
+import { execSync } from 'node:child_process'
+import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
+
+const baseDir = resolve(import.meta.dirname, '..')
+
+execSync('pnpm generate -- --total=10 --silent', {
+  cwd: baseDir,
+  stdio: 'inherit',
+})
+
+describe('Emotion build', () => {
+  test.for(['custom', 'babel', 'swc'] as const)(
+    'build:%s exits with code 0',
+    { timeout: 30_000 },
+    (name) => {
+      runBuild(name, baseDir)
+    },
+  )
+})
diff --git a/packages/emotion/benchmark/scripts/generate-app.ts b/packages/emotion/benchmark/scripts/generate-app.ts
index b4884c8..db1333a 100644
--- a/packages/emotion/benchmark/scripts/generate-app.ts
+++ b/packages/emotion/benchmark/scripts/generate-app.ts
@@ -274,7 +274,9 @@ function main() {
   mkdirSync(componentsDir, { recursive: true })
 
   const components: Array<{ type: ComponentType; index: number }> = []
-  const TOTAL = 100
+  const totalArg = process.argv.find((a) => a.startsWith('--total='))
+  const TOTAL = totalArg ? Number.parseInt(totalArg.slice('--total='.length), 10) : 100
+  const silent = process.argv.includes('--silent')
   const perType = Math.floor(TOTAL / COMPONENT_TYPES.length)
   const remainder = TOTAL % COMPONENT_TYPES.length
 
@@ -293,9 +295,11 @@ function main() {
     .join('\n')
   writeFileSync(join(componentsDir, 'index.ts'), exports + '\n')
 
-  console.log(`Generated ${components.length} components in ${componentsDir}`)
-  for (const type of COMPONENT_TYPES) {
-    console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+  if (!silent) {
+    console.log(`Generated ${components.length} components in ${componentsDir}`)
+    for (const type of COMPONENT_TYPES) {
+      console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+    }
   }
 }
 
diff --git a/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.bench.ts b/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.bench.ts
index f193c5e..10d0819 100644
--- a/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.bench.ts
+++ b/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.bench.ts
@@ -1,13 +1,18 @@
 import { bench, describe } from 'vitest'
 import { execSync } from 'node:child_process'
-import { existsSync, rmSync } from 'node:fs'
+import { existsSync, readdirSync, rmSync } from 'node:fs'
 import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
 
 const baseDir = resolve(import.meta.dirname, '..')
 const distBase = resolve(baseDir, 'dist')
 const componentsDir = resolve(baseDir, 'shared-app/src/components')
+const expectedComponents = 100
 
-if (!existsSync(componentsDir)) {
+const currentComponents = existsSync(componentsDir)
+  ? readdirSync(componentsDir).filter((f) => f.endsWith('.tsx')).length
+  : 0
+if (currentComponents !== expectedComponents) {
   execSync('pnpm generate', { cwd: baseDir, stdio: 'inherit' })
 }
 
@@ -18,18 +23,11 @@ function cleanDist(name: string) {
   }
 }
 
-function runBuild(name: string) {
-  execSync(`rolldown -c configs/${name}.ts`, {
-    cwd: baseDir,
-    stdio: 'pipe',
-  })
-}
-
 describe('JSX Remove Attributes Benchmark', () => {
   bench(
     '@rolldown/plugin-jsx-remove-attributes',
     () => {
-      runBuild('custom')
+      runBuild('custom', baseDir)
     },
     { teardown: () => cleanDist('custom') },
   )
@@ -37,7 +35,7 @@ describe('JSX Remove Attributes Benchmark', () => {
   bench(
     '@rolldown/plugin-babel',
     () => {
-      runBuild('babel')
+      runBuild('babel', baseDir)
     },
     { teardown: () => cleanDist('babel') },
   )
@@ -45,7 +43,7 @@ describe('JSX Remove Attributes Benchmark', () => {
   bench(
     '@rollup/plugin-swc',
     () => {
-      runBuild('swc')
+      runBuild('swc', baseDir)
     },
     { teardown: () => cleanDist('swc') },
   )
diff --git a/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.test.ts b/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.test.ts
new file mode 100644
index 0000000..f022749
--- /dev/null
+++ b/packages/jsx-remove-attributes/benchmark/bench/jsx-remove-attributes.test.ts
@@ -0,0 +1,21 @@
+import { describe, test } from 'vitest'
+import { execSync } from 'node:child_process'
+import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
+
+const baseDir = resolve(import.meta.dirname, '..')
+
+execSync('pnpm generate -- --total=10 --silent', {
+  cwd: baseDir,
+  stdio: 'inherit',
+})
+
+describe('JSX Remove Attributes build', () => {
+  test.for(['custom', 'babel', 'swc'] as const)(
+    'build:%s exits with code 0',
+    { timeout: 30_000 },
+    (name) => {
+      runBuild(name, baseDir)
+    },
+  )
+})
diff --git a/packages/jsx-remove-attributes/benchmark/scripts/generate-app.ts b/packages/jsx-remove-attributes/benchmark/scripts/generate-app.ts
index d77da0a..51f97fb 100644
--- a/packages/jsx-remove-attributes/benchmark/scripts/generate-app.ts
+++ b/packages/jsx-remove-attributes/benchmark/scripts/generate-app.ts
@@ -184,7 +184,9 @@ function main() {
   mkdirSync(componentsDir, { recursive: true })
 
   const components: Array<{ type: ComponentType; index: number }> = []
-  const TOTAL = 100
+  const totalArg = process.argv.find((a) => a.startsWith('--total='))
+  const TOTAL = totalArg ? Number.parseInt(totalArg.slice('--total='.length), 10) : 100
+  const silent = process.argv.includes('--silent')
   const perType = Math.floor(TOTAL / COMPONENT_TYPES.length)
   const remainder = TOTAL % COMPONENT_TYPES.length
 
@@ -203,9 +205,11 @@ function main() {
     .join('\n')
   writeFileSync(join(componentsDir, 'index.ts'), exports + '\n')
 
-  console.log(`Generated ${components.length} components in ${componentsDir}`)
-  for (const type of COMPONENT_TYPES) {
-    console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+  if (!silent) {
+    console.log(`Generated ${components.length} components in ${componentsDir}`)
+    for (const type of COMPONENT_TYPES) {
+      console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+    }
   }
 }
 
diff --git a/packages/styled-jsx/benchmark/bench/styled-jsx.bench.ts b/packages/styled-jsx/benchmark/bench/styled-jsx.bench.ts
index 44f8458..1b2a118 100644
--- a/packages/styled-jsx/benchmark/bench/styled-jsx.bench.ts
+++ b/packages/styled-jsx/benchmark/bench/styled-jsx.bench.ts
@@ -1,13 +1,18 @@
 import { bench, describe } from 'vitest'
 import { execSync } from 'node:child_process'
-import { existsSync, rmSync } from 'node:fs'
+import { existsSync, readdirSync, rmSync } from 'node:fs'
 import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
 
 const baseDir = resolve(import.meta.dirname, '..')
 const distBase = resolve(baseDir, 'dist')
 const componentsDir = resolve(baseDir, 'shared-app/src/components')
+const expectedComponents = 100
 
-if (!existsSync(componentsDir)) {
+const currentComponents = existsSync(componentsDir)
+  ? readdirSync(componentsDir).filter((f) => f.endsWith('.tsx')).length
+  : 0
+if (currentComponents !== expectedComponents) {
   execSync('pnpm generate', { cwd: baseDir, stdio: 'inherit' })
 }
 
@@ -18,18 +23,11 @@ function cleanDist(name: string) {
   }
 }
 
-function runBuild(name: string) {
-  execSync(`rolldown -c configs/${name}.ts`, {
-    cwd: baseDir,
-    stdio: 'pipe',
-  })
-}
-
 describe('Styled JSX Benchmark', () => {
   bench(
     '@rolldown/plugin-styled-jsx',
     () => {
-      runBuild('custom')
+      runBuild('custom', baseDir)
     },
     { teardown: () => cleanDist('custom') },
   )
@@ -37,7 +35,7 @@ describe('Styled JSX Benchmark', () => {
   bench(
     '@rolldown/plugin-babel',
     () => {
-      runBuild('babel')
+      runBuild('babel', baseDir)
     },
     { teardown: () => cleanDist('babel') },
   )
@@ -45,7 +43,7 @@ describe('Styled JSX Benchmark', () => {
   bench(
     '@rollup/plugin-swc',
     () => {
-      runBuild('swc')
+      runBuild('swc', baseDir)
     },
     { teardown: () => cleanDist('swc') },
   )
diff --git a/packages/styled-jsx/benchmark/bench/styled-jsx.test.ts b/packages/styled-jsx/benchmark/bench/styled-jsx.test.ts
new file mode 100644
index 0000000..e8f7754
--- /dev/null
+++ b/packages/styled-jsx/benchmark/bench/styled-jsx.test.ts
@@ -0,0 +1,21 @@
+import { describe, test } from 'vitest'
+import { execSync } from 'node:child_process'
+import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
+
+const baseDir = resolve(import.meta.dirname, '..')
+
+execSync('pnpm generate -- --total=10 --silent', {
+  cwd: baseDir,
+  stdio: 'inherit',
+})
+
+describe('Styled JSX build', () => {
+  test.for(['custom', 'babel', 'swc'] as const)(
+    'build:%s exits with code 0',
+    { timeout: 30_000 },
+    (name) => {
+      runBuild(name, baseDir)
+    },
+  )
+})
diff --git a/packages/styled-jsx/benchmark/scripts/generate-app.ts b/packages/styled-jsx/benchmark/scripts/generate-app.ts
index 49e5d3a..d35b598 100644
--- a/packages/styled-jsx/benchmark/scripts/generate-app.ts
+++ b/packages/styled-jsx/benchmark/scripts/generate-app.ts
@@ -275,7 +275,9 @@ function main() {
   mkdirSync(componentsDir, { recursive: true })
 
   const components: Array<{ type: ComponentType; index: number }> = []
-  const TOTAL = 100
+  const totalArg = process.argv.find((a) => a.startsWith('--total='))
+  const TOTAL = totalArg ? Number.parseInt(totalArg.slice('--total='.length), 10) : 100
+  const silent = process.argv.includes('--silent')
   const perType = Math.floor(TOTAL / COMPONENT_TYPES.length)
   const remainder = TOTAL % COMPONENT_TYPES.length
 
@@ -294,9 +296,11 @@ function main() {
     .join('\n')
   writeFileSync(join(componentsDir, 'index.ts'), exports + '\n')
 
-  console.log(`Generated ${components.length} components in ${componentsDir}`)
-  for (const type of COMPONENT_TYPES) {
-    console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+  if (!silent) {
+    console.log(`Generated ${components.length} components in ${componentsDir}`)
+    for (const type of COMPONENT_TYPES) {
+      console.log(`  ${type}: ${components.filter((c) => c.type === type).length}`)
+    }
   }
 }
 
diff --git a/packages/transform-imports/benchmark/bench/transform-imports.bench.ts b/packages/transform-imports/benchmark/bench/transform-imports.bench.ts
index 352d5e1..c231c99 100644
--- a/packages/transform-imports/benchmark/bench/transform-imports.bench.ts
+++ b/packages/transform-imports/benchmark/bench/transform-imports.bench.ts
@@ -1,13 +1,18 @@
 import { bench, describe } from 'vitest'
 import { execSync } from 'node:child_process'
-import { existsSync, rmSync } from 'node:fs'
+import { existsSync, readdirSync, rmSync } from 'node:fs'
 import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
 
 const baseDir = resolve(import.meta.dirname, '..')
 const distBase = resolve(baseDir, 'dist')
 const modulesDir = resolve(baseDir, 'shared-app/src/modules')
+const expectedModules = 100
 
-if (!existsSync(modulesDir)) {
+const currentModules = existsSync(modulesDir)
+  ? readdirSync(modulesDir).filter((f) => f.endsWith('.js')).length
+  : 0
+if (currentModules !== expectedModules) {
   execSync('pnpm generate', { cwd: baseDir, stdio: 'inherit' })
 }
 
@@ -18,18 +23,11 @@ function cleanDist(name: string) {
   }
 }
 
-function runBuild(name: string) {
-  execSync(`rolldown -c configs/${name}.ts`, {
-    cwd: baseDir,
-    stdio: 'pipe',
-  })
-}
-
 describe('Transform Imports Benchmark', () => {
   bench(
     '@rolldown/plugin-transform-imports',
     () => {
-      runBuild('custom')
+      runBuild('custom', baseDir)
     },
     { teardown: () => cleanDist('custom') },
   )
@@ -37,7 +35,7 @@ describe('Transform Imports Benchmark', () => {
   bench(
     'babel-plugin-transform-imports',
     () => {
-      runBuild('babel')
+      runBuild('babel', baseDir)
     },
     { teardown: () => cleanDist('babel') },
   )
@@ -45,7 +43,7 @@ describe('Transform Imports Benchmark', () => {
   bench(
     '@swc/plugin-transform-imports',
     () => {
-      runBuild('swc')
+      runBuild('swc', baseDir)
     },
     { teardown: () => cleanDist('swc') },
   )
diff --git a/packages/transform-imports/benchmark/bench/transform-imports.test.ts b/packages/transform-imports/benchmark/bench/transform-imports.test.ts
new file mode 100644
index 0000000..c1d136e
--- /dev/null
+++ b/packages/transform-imports/benchmark/bench/transform-imports.test.ts
@@ -0,0 +1,21 @@
+import { describe, test } from 'vitest'
+import { execSync } from 'node:child_process'
+import { resolve } from 'node:path'
+import { runBuild } from '@rolldown/benchmark-utils/run-build'
+
+const baseDir = resolve(import.meta.dirname, '..')
+
+execSync('pnpm generate -- --total=10 --silent', {
+  cwd: baseDir,
+  stdio: 'inherit',
+})
+
+describe('Transform Imports build', () => {
+  test.for(['custom', 'babel', 'swc'] as const)(
+    'build:%s exits with code 0',
+    { timeout: 30_000 },
+    (name) => {
+      runBuild(name, baseDir)
+    },
+  )
+})
diff --git a/packages/transform-imports/benchmark/scripts/generate-app.ts b/packages/transform-imports/benchmark/scripts/generate-app.ts
index 59c8d51..c48bad5 100644
--- a/packages/transform-imports/benchmark/scripts/generate-app.ts
+++ b/packages/transform-imports/benchmark/scripts/generate-app.ts
@@ -166,7 +166,9 @@ function main() {
   if (existsSync(appDir)) rmSync(appDir, { recursive: true })
   mkdirSync(join(appDir, 'modules'), { recursive: true })
 
-  const TOTAL = 100
+  const totalArg = process.argv.find((a) => a.startsWith('--total='))
+  const TOTAL = totalArg ? Number.parseInt(totalArg.slice('--total='.length), 10) : 100
+  const silent = process.argv.includes('--silent')
   const files: string[] = []
 
   for (let i = 0; i < TOTAL; i++) {
@@ -185,9 +187,11 @@ function main() {
     .join('\n')
   writeFileSync(join(appDir, 'index.js'), entry + '\n')
 
-  console.log(`Generated ${TOTAL} modules in ${appDir}/modules/`)
-  for (const type of PATTERN_TYPES) {
-    console.log(`  Pattern distribution includes: ${type}`)
+  if (!silent) {
+    console.log(`Generated ${TOTAL} modules in ${appDir}/modules/`)
+    for (const type of PATTERN_TYPES) {
+      console.log(`  Pattern distribution includes: ${type}`)
+    }
   }
 }