@tangle-network/agent-bench 0.9.3 → 0.9.4

package/CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## 0.9.4
+
+Requires Runtime 0.202.0 so Bench consumers receive explicit failed-execution settlement and retained failure recovery.
+Supports an optional Node test concurrency limit for hosts with constrained memory.
+
 ## 0.9.3
 
 Requires Runtime 0.201.0, Interface 2.5.0, Eval 0.176, and Knowledge 15 for scoped candidate knowledge.

package/HARNESS.md

@@ -37,6 +37,14 @@ pnpm verify:bench
 pnpm verify:official-optimizers
 ```
 
+Set `AGENT_BENCH_PACKAGE_TEST_CONCURRENCY=1` to run Node source test files sequentially on memory-constrained hosts.
+The value must be a positive safe integer; leaving it unset preserves Node's default concurrency.
+Vitest uses the worker limit in `bench/vitest.config.ts`.
+
+Packing resolves the Runtime dependency from `workspace:^` to a version range.
+A new Runtime compatibility line does not update previously published Bench packages.
+Check the packed dependency range and release Bench when consumers must receive that Runtime version.
+
 `verify:official-optimizers` exercises the official Optimize Anything bridge, engine identities, equal input budgets, resume compatibility, candidate callbacks, accounting, and package provenance. Its deterministic candidate improvement is deliberately a fixture. It does **not** reproduce the published GEPA or Omni benchmark numbers.
 
 ### Bounded benchmark matrix

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-bench",
-  "version": "0.9.3",
+  "version": "0.9.4",
   "type": "module",
   "description": "Benchmark adapters and execution for agent-runtime across coding, tool-use, RAG, memory, browser, and terminal tasks.",
   "repository": {
@@ -29,7 +29,7 @@
     "@tangle-network/agent-interface": "^2.5.0",
     "@tangle-network/agent-knowledge": "^15.0.0",
     "@tangle-network/sandbox": ">=0.36.4 <0.38.0",
-    "@tangle-network/agent-runtime": "^0.201.0"
+    "@tangle-network/agent-runtime": "^0.202.0"
   },
   "devDependencies": {
     "@arethetypeswrong/cli": "0.18.5",

package/scripts/run-package-tests.mjs

@@ -8,6 +8,21 @@ const execFileAsync = promisify(execFile)
 const benchDir = path.resolve(path.dirname(fileURLToPath(import.meta.url)), '..')
 const sourceDir = path.join(benchDir, 'src')
 
+export function packageNodeTestArgs(files, env = process.env) {
+  const raw = env.AGENT_BENCH_PACKAGE_TEST_CONCURRENCY
+  const concurrency = raw === undefined ? undefined : Number(raw)
+  if (raw !== undefined && (!/^[1-9]\d*$/.test(raw) || !Number.isSafeInteger(concurrency))) {
+    throw new Error('AGENT_BENCH_PACKAGE_TEST_CONCURRENCY must be a positive safe integer')
+  }
+  return [
+    '--test',
+    ...(concurrency === undefined ? [] : [`--test-concurrency=${concurrency}`]),
+    '--import',
+    'tsx',
+    ...files,
+  ]
+}
+
 export function resolvePackageTestTimeoutMs(env = process.env) {
   const raw = env.AGENT_BENCH_PACKAGE_TEST_TIMEOUT_MS
   if (raw === undefined) return undefined
@@ -74,7 +89,7 @@ async function main() {
   if (nodeTests.length > 0) {
     await run(
       process.execPath,
-      ['--test', '--import', 'tsx', ...nodeTests.map((file) => path.relative(benchDir, file))],
+      packageNodeTestArgs(nodeTests.map((file) => path.relative(benchDir, file))),
       {
         ...process.env,
         TSX_TSCONFIG_PATH: 'tsconfig.public.json',

package/scripts/run-package-tests.test.mjs

@@ -1,6 +1,20 @@
 import assert from 'node:assert/strict'
 import { test } from 'node:test'
-import { resolvePackageTestTimeoutMs, run } from './run-package-tests.mjs'
+import { packageNodeTestArgs, resolvePackageTestTimeoutMs, run } from './run-package-tests.mjs'
+
+test('package test concurrency reaches Node without changing selected files', () => {
+  const files = ['src/first.test.mts', 'src/second.test.ts']
+  assert.deepEqual(packageNodeTestArgs(files, {}), ['--test', '--import', 'tsx', ...files])
+  assert.deepEqual(packageNodeTestArgs(files, { AGENT_BENCH_PACKAGE_TEST_CONCURRENCY: '1' }), [
+    '--test', '--test-concurrency=1', '--import', 'tsx', ...files,
+  ])
+  for (const value of ['', '0', '-1', '1.5', 'Infinity', '9007199254740992']) {
+    assert.throws(
+      () => packageNodeTestArgs(files, { AGENT_BENCH_PACKAGE_TEST_CONCURRENCY: value }),
+      /must be a positive safe integer/,
+    )
+  }
+})
 
 test('package test timeout is optional and caller-controlled', () => {
   assert.equal(resolvePackageTestTimeoutMs({}), undefined)