diff --git a/scripts/lib/shard-engine.ts b/scripts/lib/shard-engine.ts
index a0b3bfa03..341d8a3b6 100644
--- a/scripts/lib/shard-engine.ts
+++ b/scripts/lib/shard-engine.ts
@@ -12,6 +12,8 @@
 
 import { spawn, type ChildProcess } from 'node:child_process';
 import { StringDecoder } from 'node:string_decoder';
+import * as fs from 'node:fs';
+import * as os from 'node:os';
 import * as path from 'node:path';
 
 const ROOT = path.resolve(import.meta.dir, '..', '..');
@@ -402,6 +404,18 @@ export interface RunShardChildOptions {
    * return reports incomplete capture instead of treating the prefix as final.
    */
   hookStreams: (child: ChildProcess) => Array<Promise<void>>;
+  /**
+   * Lane-owned companions of the child (the free lane's detached-browser
+   * tracker). Created right after spawn; `signal` runs after every forwarded
+   * group kill, and `settle` runs after the final group kill while parent
+   * signals are still forwarded.
+   */
+  attach?: (child: ChildProcess) => ShardChildCompanion;
+}
+
+export interface ShardChildCompanion {
+  signal(force: boolean): void;
+  settle(): Promise<void>;
 }
 
 export interface ShardChildResult {
@@ -436,6 +450,8 @@ export interface ShardChildResult {
  * call: a spawn 'error' event THROWS from here after the finally block runs,
  * preserving the runners' existing could-not-run handling.
  */
+const REAP_GRACE_MS = 250;
+
 export async function runShardChild(options: RunShardChildOptions): Promise<ShardChildResult> {
   const deadlineMs = Math.min(options.deadlineMs ?? Infinity, Date.now() + options.timeoutMs);
   if (!Number.isFinite(deadlineMs)) throw new Error('Shard deadline must be finite');
@@ -451,10 +467,12 @@ export async function runShardChild(options: RunShardChildOptions): Promise<Shar
     windowsHide: true,
   });
   const groupPid = child.pid ?? null;
+  const companion = options.attach?.(child);
   // Group-kill on parent SIGINT/SIGTERM too, not just on timeout.
   const forwarding = installChildSignalForwarding({
     kill: (signal?: NodeJS.Signals | number) => {
       killProcessGroup(child, (signal as NodeJS.Signals) ?? 'SIGTERM');
+      companion?.signal(signal === 'SIGKILL');
       return true;
     },
   });
@@ -477,8 +495,10 @@ export async function runShardChild(options: RunShardChildOptions): Promise<Shar
   };
   let close!: () => void;
   const closed = new Promise<void>(resolve => { close = resolve; });
-  const onExit = (code: number | null) => { exitCode = code; };
-  const onClose = (code: number | null) => { exitCode = code; childClosed = true; close(); };
+  let reap!: () => void;
+  const reaped = new Promise<void>(resolve => { reap = resolve; });
+  const onExit = (code: number | null) => { exitCode = code; reap(); };
+  const onClose = (code: number | null) => { exitCode = code; childClosed = true; close(); reap(); };
   child.once('exit', onExit);
   child.once('close', onClose);
   child.on('error', rememberError);
@@ -506,10 +526,19 @@ export async function runShardChild(options: RunShardChildOptions): Promise<Shar
     )));
     await Promise.race([Promise.all([closed, drainage]), expired]);
     if (Date.now() >= deadlineMs) timedOut = true;
+    // A wall-killed child is normally reaped within milliseconds; wait that
+    // long (bounded) so callers never observe a live pid after a timeout.
+    let reapTimer: ReturnType<typeof setTimeout> | undefined;
+    await Promise.race([reaped, new Promise<void>(resolve => { reapTimer = setTimeout(resolve, REAP_GRACE_MS); })]);
+    clearTimeout(reapTimer);
   } finally {
     clearTimeout(killTimer);
-    forwarding.dispose();
     kill();
+    if (companion) {
+      try { await companion.settle(); }
+      catch (error) { rememberError(error); }
+    }
+    forwarding.dispose();
     child.off('exit', onExit);
     child.off('close', onClose);
     if (!childClosed || pendingStreams > 0) {
@@ -530,3 +559,157 @@ export async function runShardChild(options: RunShardChildOptions): Promise<Shar
   }
   return result;
 }
+
+// --- Per-shard sandbox, logs, duration seeds, verdicts, CLI flags ---
+
+/**
+ * Per-shard temp + Chromium-profile isolation. Two concurrent shards on one
+ * profile dir kill each other's browser, and shared tmp cross-contaminates;
+ * a group-SIGKILLed shard never runs its own cleanup, so the lane removes
+ * `stateDir` afterwards (the cleanup backstop). `realpath` resolves a
+ * symlinked tmpdir (macOS /var -> /private/var) for lanes that compare paths.
+ */
+export function createShardSandbox(
+  prefix: string,
+  baseEnv: NodeJS.ProcessEnv,
+  options: { realpath?: boolean } = {},
+): { stateDir: string; tmp: string; env: NodeJS.ProcessEnv } {
+  const created = fs.mkdtempSync(path.join(os.tmpdir(), prefix));
+  const stateDir = options.realpath ? fs.realpathSync(created) : created;
+  const tmp = path.join(stateDir, 'tmp');
+  fs.mkdirSync(tmp);
+  const env: NodeJS.ProcessEnv = {
+    ...baseEnv, TMPDIR: tmp, TEMP: tmp, TMP: tmp,
+    CHROMIUM_PROFILE: path.join(stateDir, 'chromium-profile'),
+  };
+  return { stateDir, tmp, env };
+}
+
+/**
+ * Asynchronous best-effort backstop removal: a SIGKILLed shard can leave a
+ * full git workspace plus a Chromium profile, and a synchronous recursive
+ * delete would stall every sibling shard's classification and timers.
+ */
+export async function removeShardSandbox(stateDir: string): Promise<void> {
+  try { await fs.promises.rm(stateDir, { recursive: true, force: true }); }
+  catch { /* a locked file must not turn a real verdict into an exception */ }
+}
+
+let shardLogSequence = 0;
+
+/** Timestamped log path; pid + sequence defeat same-millisecond collisions. */
+export function nextShardLogPath(directory: string, stem: string): string {
+  const stamp = new Date().toISOString().replace(/[:.]/g, '-');
+  shardLogSequence += 1;
+  return path.join(directory, `${stem}-${stamp}-${process.pid}-${shardLogSequence}.log`);
+}
+
+export interface ShardLog {
+  readonly path: string;
+  readonly stream: fs.WriteStream;
+  /** Set on the first write error; later chunks are dropped, never thrown. */
+  failed: boolean;
+  write(chunk: Buffer | string): void;
+}
+
+/** Full-stream capture: every child byte is spooled to disk, never held in RAM. */
+export function openShardLog(logPath: string, label: string, mode?: number): ShardLog {
+  const stream = fs.createWriteStream(logPath, mode === undefined ? undefined : { mode });
+  const log: ShardLog = {
+    path: logPath, stream, failed: false,
+    write(chunk) { if (!log.failed) stream.write(chunk); },
+  };
+  stream.on('error', (err) => {
+    if (log.failed) return;
+    log.failed = true;
+    console.error(`${label} could not write the full log at ${logPath}: ${err.message}`);
+  });
+  return log;
+}
+
+export type DurationSeedRead =
+  | { status: 'missing' }
+  | { status: 'corrupt'; error: Error }
+  | { status: 'ok'; durations: Record<string, number> };
+
+/** One reader for `{ durations: { file: ms } }` seeds; the lane decides which values count. */
+export function readDurationSeed(file: string, accepts: (ms: number) => boolean): DurationSeedRead {
+  let raw: string;
+  try { raw = fs.readFileSync(file, 'utf-8'); }
+  catch { return { status: 'missing' }; }
+  try {
+    const parsed = JSON.parse(raw) as { durations?: Record<string, unknown> };
+    return { status: 'ok', durations: Object.fromEntries(Object.entries(parsed.durations ?? {})
+      .filter((entry): entry is [string, number] =>
+        typeof entry[1] === 'number' && Number.isFinite(entry[1]) && accepts(entry[1]))) };
+  } catch (error) {
+    return { status: 'corrupt', error: error as Error };
+  }
+}
+
+/** Atomic temp+rename: a killed writer never leaves a truncated seed behind. */
+export function writeDurationSeed(file: string, durations: Record<string, number>): void {
+  const payload = {
+    version: 1,
+    recordedAt: new Date().toISOString(),
+    durations: Object.fromEntries(Object.entries(durations).sort(([a], [b]) => (a < b ? -1 : 1))),
+  };
+  const temporary = `${file}.tmp-${process.pid}`;
+  fs.writeFileSync(temporary, `${JSON.stringify(payload, null, 2)}\n`);
+  fs.renameSync(temporary, file);
+}
+
+/** What a strictly passed shard that executed zero tests becomes. */
+export type ZeroExecutionVerdict = 'passed' | 'passed-with-warning' | 'passed-empty';
+
+/** Classification rules each lane injects; the engine applies them, never decides them. */
+export interface LanePolicy {
+  /** Which recorded seed durations the lane trusts (free >= 0, paid > 0). */
+  acceptsSeedDuration(ms: number): boolean;
+  /** `promisedAll`: the run promised every test (EVALS_ALL) rather than a selection. */
+  zeroExecution(run: { promisedAll: boolean }): ZeroExecutionVerdict;
+}
+
+export function zeroExecutionVerdict(
+  executedTests: number | null,
+  policy: LanePolicy,
+  run: { promisedAll: boolean },
+): ZeroExecutionVerdict {
+  return executedTests === 0 ? policy.zeroExecution(run) : 'passed';
+}
+
+export type StrictShardStatus = 'passed' | 'failed' | 'timed-out';
+
+/**
+ * The verdict both lanes share: a wall timeout is its own status; otherwise
+ * a shard passes only with complete evidence (log, capture, cleanup) AND a
+ * strict exit of zero, which requires bun's summary to count every planned file.
+ */
+export function strictShardStatus(input: {
+  timedOut: boolean;
+  evidenceComplete: boolean;
+  exitCode: number | null;
+  summary: BunTestOutputSummary;
+  expectedFiles: number;
+}): StrictShardStatus {
+  if (input.timedOut) return 'timed-out';
+  return input.evidenceComplete && strictTestExitCode(input.exitCode ?? 1, input.summary, input.expectedFiles) === 0
+    ? 'passed' : 'failed';
+}
+
+/**
+ * Shared flag loop. Each lane declares its flags; a handler that takes a
+ * value calls `next()` (the following argv entry, or undefined) and owns its
+ * own validation message. Any undeclared flag is `Unknown argument: <flag>`.
+ */
+export function parseCliFlags(
+  argv: string[],
+  handlers: Record<string, (next: () => string | undefined) => void>,
+): void {
+  for (let index = 0; index < argv.length; index += 1) {
+    const arg = argv[index];
+    const handler = Object.hasOwn(handlers, arg) ? handlers[arg] : undefined;
+    if (!handler) throw new Error(`Unknown argument: ${arg}`);
+    handler(() => argv[++index]);
+  }
+}
diff --git a/scripts/test-free-shards.ts b/scripts/test-free-shards.ts
index 1c6dcd233..445114291 100755
--- a/scripts/test-free-shards.ts
+++ b/scripts/test-free-shards.ts
@@ -11,9 +11,9 @@
  *      patterns (`/bin/bash`, `sh -c`, raw `/tmp/`, `chmod`, `xargs`). Files
  *      that match are excluded from the Windows-safe subset — they would fail
  *      on `windows-latest` no matter how the runner shards them.
- *   4. Execution. Spawn `bun test` children and refuse to trust their exit
- *      code alone: every byte of output is classified through
- *      scripts/test-strict-output.ts, so a child that exits 0 without bun's
+ *   4. Execution. Run `bun test` children through the shared shard engine
+ *      (scripts/lib/shard-engine.ts runShardChild) and refuse to trust their
+ *      exit code alone: every byte of output is classified strictly, so a child that exits 0 without bun's
  *      terminal summary (a mid-suite process.exit truncation), with `(fail)`
  *      result lines, or with fewer files run than planned is a FAILURE. An
  *      external wall-clock timeout SIGKILLs the child's process group and
@@ -87,18 +87,38 @@ import { createHash, randomUUID } from 'node:crypto';
 import { isPaidTestFile } from '../test/helpers/paid-test-set';
 import {
   BunTestOutputClassifier,
+  createShardSandbox,
   exactTestFileSelectors,
-  installChildSignalForwarding,
   isTerminationRequested,
   killProcessGroup,
   BunFailureSummaryParser,
+  nextShardLogPath,
+  openShardLog,
   parseBunFailureResult,
+  parseCliFlags,
   normalizeRelativePath,
-  strictTestExitCode,
+  readDurationSeed,
+  runShardChild,
+  strictShardStatus,
   stripAnsiLine,
-} from './test-strict-output';
+  writeDurationSeed,
+  zeroExecutionVerdict,
+  type LanePolicy,
+  type ShardChildResult,
+} from './lib/shard-engine';
 
-export { normalizeRelativePath } from './test-strict-output';
+export { normalizeRelativePath } from './lib/shard-engine';
+
+/**
+ * Free-lane classification policy. Seeds accept zero-duration files (fast
+ * files are real measurements). A shard with zero executed tests passes when
+ * bun's summary still counted every planned file; missing or short file
+ * counts already fail through the strict verdict.
+ */
+export const FREE_LANE_POLICY: LanePolicy = {
+  acceptsSeedDuration: (ms) => ms >= 0,
+  zeroExecution: () => 'passed',
+};
 
 const ROOT = path.resolve(import.meta.dir, '..');
 // design/test was silently absent from BOTH the package.json test script and
@@ -669,24 +689,15 @@ export const FREE_TEST_DURATIONS_FILE = 'scripts/free-test-durations.json';
 export function loadFreeTestDurations(rootDir = ROOT): Record<string, number> | null {
   const file = process.env.GSTACK_FREE_TEST_DURATIONS
     ?? path.join(rootDir, FREE_TEST_DURATIONS_FILE);
-  let raw: string;
-  try {
-    raw = fs.readFileSync(file, 'utf-8');
-  } catch {
-    return null; // no seed — hash sharding, silently (fresh checkouts are normal)
-  }
-  try {
-    const parsed = JSON.parse(raw) as { durations?: Record<string, unknown> };
-    const entries = Object.entries(parsed.durations ?? {})
-      .filter((entry): entry is [string, number] =>
-        typeof entry[1] === 'number' && Number.isFinite(entry[1]) && entry[1] >= 0);
-    if (entries.length === 0) return null;
-    return Object.fromEntries(entries);
-  } catch (error) {
+  const seed = readDurationSeed(file, FREE_LANE_POLICY.acceptsSeedDuration);
+  // No seed — hash sharding, silently (fresh checkouts are normal).
+  if (seed.status === 'missing') return null;
+  if (seed.status === 'corrupt') {
     // A corrupt seed (bad merge) must cost a warning, never the suite.
-    console.error(`[test:free] WARNING: corrupt durations seed ${file} (${(error as Error).message}) — falling back to hash sharding`);
+    console.error(`[test:free] WARNING: corrupt durations seed ${file} (${seed.error.message}) — falling back to hash sharding`);
     return null;
   }
+  return Object.keys(seed.durations).length === 0 ? null : seed.durations;
 }
 
 export interface PackedShards {
@@ -889,7 +900,7 @@ type CliOptions = {
   results: string | null;
 };
 
-function parseCliOptions(argv: string[]): CliOptions {
+export function parseCliOptions(argv: string[]): CliOptions {
   let dryRun = false;
   let listOnly = false;
   let recordDurations = false;
@@ -903,46 +914,41 @@ function parseCliOptions(argv: string[]): CliOptions {
   const paths: Record<'ciPlan' | 'ciRun' | 'ciVerify' | 'result' | 'results', string | null> = {
     ciPlan: null, ciRun: null, ciVerify: null, result: null, results: null,
   };
+  const pathFlag = (flag: string, key: keyof typeof paths) => (next: () => string | undefined) => {
+    const value = next();
+    if (!value || value.startsWith('--')) throw new Error(`Missing path for ${flag}`);
+    paths[key] = value;
+  };
 
-  for (let index = 0; index < argv.length; index += 1) {
-    const arg = argv[index];
-    if (arg === '--dry-run') { dryRun = true; continue; }
-    if (arg === '--list') { listOnly = true; continue; }
-    if (arg === '--record-durations') { recordDurations = true; continue; }
-    if (arg === '--windows-only') { windowsOnly = true; continue; }
-    if (arg === '--verbose') { verbose = true; continue; }
-    if (arg === '--quick') { quick = true; continue; }
-    const pathKey = ({ '--ci-plan': 'ciPlan', '--ci-run': 'ciRun', '--ci-verify': 'ciVerify', '--result': 'result', '--results': 'results' } as const)[arg];
-    if (pathKey) {
-      const value = argv[++index];
-      if (!value || value.startsWith('--')) throw new Error(`Missing path for ${arg}`);
-      paths[pathKey] = value;
-      continue;
-    }
-    if (arg === '--shards') {
-      const value = argv[index + 1];
+  parseCliFlags(argv, {
+    '--dry-run': () => { dryRun = true; },
+    '--list': () => { listOnly = true; },
+    '--record-durations': () => { recordDurations = true; },
+    '--windows-only': () => { windowsOnly = true; },
+    '--verbose': () => { verbose = true; },
+    '--quick': () => { quick = true; },
+    '--ci-plan': pathFlag('--ci-plan', 'ciPlan'),
+    '--ci-run': pathFlag('--ci-run', 'ciRun'),
+    '--ci-verify': pathFlag('--ci-verify', 'ciVerify'),
+    '--result': pathFlag('--result', 'result'),
+    '--results': pathFlag('--results', 'results'),
+    '--shards': (next) => {
+      const value = next();
       if (!value) throw new Error('Missing value for --shards');
       shardCount = Number.parseInt(value, 10);
-      index += 1;
-      continue;
-    }
-    if (arg === '--shard') {
-      const value = argv[index + 1];
+    },
+    '--shard': (next) => {
+      const value = next();
       if (!value) throw new Error('Missing value for --shard');
       shardIndex = Number.parseInt(value, 10);
-      index += 1;
-      continue;
-    }
-    if (arg === '--wall-timeout') {
-      const value = Number.parseInt(argv[index + 1] ?? '', 10);
+    },
+    '--wall-timeout': (next) => {
+      const value = Number.parseInt(next() ?? '', 10);
       if (!Number.isInteger(value) || value <= 0) throw new Error('--wall-timeout needs a positive integer (seconds)');
       wallTimeoutMs = value * 1000;
       wallTimeoutExplicit = true;
-      index += 1;
-      continue;
-    }
-    throw new Error(`Unknown argument: ${arg}`);
-  }
+    },
+  });
 
   const ciModes = [paths.ciPlan, paths.ciRun, paths.ciVerify].filter(Boolean).length;
   if (ciModes > 1 || (ciModes && (quick || listOnly || dryRun || recordDurations || windowsOnly))) throw new Error('CI modes cannot be combined with other selection modes');
@@ -1807,63 +1813,23 @@ export async function runFreeShard(
   // shows. Printed once at start so a wedged or noisy run is inspectable
   // without a re-run.
   const logPath = options.logFilePath ?? nextDefaultLogPath(rootDir);
-  const logStream = fs.createWriteStream(logPath, { mode: 0o600 });
-  let logWriteFailed = false;
-  logStream.on('error', (err) => {
-    if (logWriteFailed) return;
-    logWriteFailed = true;
-    console.error(`${label} could not write the full log at ${logPath}: ${err.message}`);
-  });
+  const shardLog = openShardLog(logPath, label, 0o600);
   log(`[test:free] full log: ${logPath}`);
 
   const { command, args } = options.commandFor
     ? options.commandFor(files)
     : { command: process.execPath, args: buildShardArgs(files, { parallel: options.parallel, rootDir }) };
 
-  const env = { ...(options.env ?? process.env) };
-  const stateDir = fs.realpathSync(fs.mkdtempSync(path.join(os.tmpdir(), 'gstack-free-shard-')));
-  const childTmp = path.join(stateDir, 'tmp');
-  fs.mkdirSync(childTmp);
-  env.TMPDIR = childTmp;
-  env.TEMP = childTmp;
-  env.TMP = childTmp;
+  // realpath: the browser tracker refuses a state dir whose path resolves elsewhere.
+  const { stateDir, env } = createShardSandbox('gstack-free-shard-', options.env ?? process.env, { realpath: true });
   // CLI renders otherwise attach to the repo's shared .gstack/browse.json,
   // even with distinct Chromium profiles. Concurrent shards and surviving
   // daemons from prior runs can then replace or remove each other's state.
   // Override inherited state too; the shard owns this directory's cleanup.
   env.BROWSE_STATE_FILE = path.join(stateDir, '.gstack', 'browse.json');
   env.GSTACK_FREE_SHARD_ID = randomUUID();
-  // Per-shard Chromium profile (same isolation idea as TMPDIR): nine test
-  // files launch in-process persistent contexts or daemons that default to
-  // the SHARED ~/.gstack/chromium-profile, and two concurrent shards on one
-  // profile dir kill each other's browser — observed live on CI once
-  // duration packing recomposed shards (handoff's launchPersistentContext
-  // died "Target page, context or browser has been closed" while a sibling
-  // shard's daemon logged "Chromium process crashed"). Hash sharding had
-  // masked the collision by chance placement. Within a shard, files run
-  // serially, so sharing the per-shard profile is safe; config tests that
-  // assert resolution order save/restore this env around their assertions.
-  env.CHROMIUM_PROFILE = path.join(stateDir, 'chromium-profile');
 
   const startedAt = Date.now();
-  const child = spawn(command, args, {
-    cwd: rootDir,
-    env,
-    stdio: ['ignore', 'pipe', 'pipe'],
-    detached: process.platform !== 'win32',
-    windowsHide: true,
-  });
-  const groupPid = child.pid ?? null;
-  const browser = trackShardBrowser(stateDir, env);
-  // Group-kill on parent SIGINT/SIGTERM too, not just on timeout.
-  const forwarding = installChildSignalForwarding({
-    kill: (signal?: NodeJS.Signals | number) => {
-      killProcessGroup(child, (signal as NodeJS.Signals) ?? 'SIGTERM');
-      browser.signal(signal === 'SIGKILL');
-      return true;
-    },
-  });
-
   const classifier = new BunTestOutputClassifier();
 
   // Console policy: quiet => nothing; verbose => the raw firehose; default =>
@@ -1881,14 +1847,16 @@ export async function runFreeShard(
   const reporter = new FreeRunReporter(files, options.verbose ? undefined : emitToConsole);
 
   const captureFailures = new Map<StreamOrigin, Error>();
-  const consumeStream = (stream: typeof child.stdout, origin: StreamOrigin): Promise<void> =>
-    new Promise((resolve) => {
+  const drained = new Set<StreamOrigin>();
+  const consumeStream = (stream: NodeJS.ReadableStream | null, origin: StreamOrigin): Promise<void> =>
+    new Promise<void>((resolve) => {
       if (!stream) {
         captureFailures.set(origin, new Error('configured pipe is missing'));
         resolve();
         return;
       }
-      let ended = stream.readableEnded;
+      const readable = stream as NodeJS.ReadableStream & { readableEnded: boolean; destroyed: boolean; errored: Error | null };
+      let ended = readable.readableEnded;
       const incomplete = (error?: Error | null): void => {
         // A delayed error replaces the initial destroyed-stream diagnostic
         // with its original cause. Failures are data, never early rejections
@@ -1901,47 +1869,49 @@ export async function runFreeShard(
       stream.on('error', incomplete);
       stream.once('end', () => { ended = true; resolve(); });
       stream.once('close', () => {
-        if (!ended) incomplete(stream.errored);
+        if (!ended) incomplete(readable.errored);
         else resolve();
       });
       stream.on('data', (chunk: Buffer | string) => {
         classifier.write(chunk, origin); // strict verdict ALWAYS sees the full stream
-        if (!logWriteFailed) logStream.write(chunk);
+        shardLog.write(chunk);
         reporter.write(chunk, origin);
         if (options.verbose) emitToConsole(typeof chunk === 'string' ? chunk : chunk.toString('utf8'), origin);
       });
       if (ended) resolve();
-      else if (stream.destroyed) incomplete(stream.errored);
-    });
+      else if (readable.destroyed) incomplete(readable.errored);
+    }).then(() => { drained.add(origin); });
 
-  let timedOut = false;
-  const killTimer = setTimeout(() => {
-    timedOut = true;
-    killProcessGroup(child, 'SIGKILL');
-  }, wallTimeoutMs);
-
-  let exitCode: number | null = null;
-  let cleanupError: string | null = null;
+  let child: ShardChildResult = { exitCode: null, timedOut: false, groupPid: null };
+  let cleanupError = null as string | null;
   try {
-    const streams = [consumeStream(child.stdout, 'stdout'), consumeStream(child.stderr, 'stderr')];
-    exitCode = await new Promise<number | null>((resolve, reject) => {
-      child.once('error', reject);
-      child.once('close', (code) => resolve(code));
+    // Shared spawn/detached/group-kill/wall-timer/reap lifecycle. The browser
+    // tracker rides along: forwarded signals reach it, and it settles after
+    // the final group kill.
+    child = await runShardChild({
+      command, args, cwd: rootDir, env, timeoutMs: wallTimeoutMs,
+      attach: () => {
+        const browser = trackShardBrowser(stateDir, env);
+        return { signal: (force) => browser.signal(force), settle: async () => { cleanupError = await browser.settle(); } };
+      },
+      hookStreams: (spawned) => [consumeStream(spawned.stdout, 'stdout'), consumeStream(spawned.stderr, 'stderr')],
     });
-    await Promise.all(streams);
+  } catch (error) {
+    child = (error as { shardResult?: ShardChildResult } | null)?.shardResult ?? child;
+    throw error;
   } finally {
-    clearTimeout(killTimer);
-    // Reap survivors of this shard even on the clean path.
-    killProcessGroup(child, 'SIGKILL');
-    cleanupError = await browser.settle();
+    // A wall-expired child is not drained: its unread tail is lost evidence.
+    for (const origin of ['stdout', 'stderr'] as const) {
+      if (!drained.has(origin) && !captureFailures.has(origin)) captureFailures.set(origin, new Error('stream did not drain before the wall deadline'));
+    }
     reporter.end();
     for (const [origin, error] of captureFailures) {
       const diagnostic = `${label} ${origin} capture incomplete: ${error.message} `
-        + `(child exit ${exitCode ?? 'signal'}). Full log: ${logPath}`;
+        + `(child exit ${child.exitCode ?? 'signal'}). Full log: ${logPath}`;
       console.error(diagnostic);
-      if (!logWriteFailed) logStream.write(diagnostic + '\n');
+      shardLog.write(diagnostic + '\n');
     }
-    await new Promise<void>((resolve) => logStream.end(() => resolve()));
+    await new Promise<void>((resolve) => shardLog.stream.end(() => resolve()));
     try {
       if (!cleanupError) fs.rmSync(stateDir, { recursive: true, force: true });
     } catch {
@@ -1949,13 +1919,18 @@ export async function runFreeShard(
       // Windows must not turn a real verdict into an exception.
       if (process.platform !== 'win32') cleanupError = 'could not remove the owned shard directory';
     }
-    forwarding.dispose();
   }
 
+  const { exitCode, timedOut, groupPid } = child;
+  const logWriteFailed = shardLog.failed;
   const summary = classifier.end();
-  const status: FreeShardStatus = timedOut
-    ? 'timed-out'
-    : !cleanupError && !logWriteFailed && captureFailures.size === 0 && strictTestExitCode(exitCode ?? 1, summary, files.length) === 0 ? 'passed' : 'failed';
+  let status: FreeShardStatus = strictShardStatus({
+    timedOut, exitCode, summary, expectedFiles: files.length,
+    evidenceComplete: !cleanupError && !logWriteFailed && captureFailures.size === 0,
+  });
+  if (status === 'passed' && zeroExecutionVerdict(reporter.report().testsRan, FREE_LANE_POLICY, { promisedAll: true }) === 'passed-empty') {
+    status = 'failed';
+  }
 
   if (cleanupError) console.error(`${label} browser cleanup failed: ${cleanupError}; retained ${stateDir}`);
 
@@ -2011,9 +1986,7 @@ export async function runFreeShard(
   return outcome;
 }
 
-let logPathSequence = 0;
-
-/** Timestamped retained log file; pid+sequence defeat same-ms collisions. */
+/** Retained private log under .context/free-test-logs; never through a link. */
 function nextDefaultLogPath(rootDir: string): string {
   let directory = fs.realpathSync(rootDir);
   for (const part of ['.context', 'free-test-logs']) {
@@ -2023,12 +1996,10 @@ function nextDefaultLogPath(rootDir: string): string {
     if (!existing) fs.mkdirSync(directory, { mode: 0o700 });
   }
   fs.chmodSync(directory, 0o700);
-  const stamp = new Date().toISOString().replace(/[:.]/g, '-');
-  logPathSequence += 1;
-  return path.join(directory, `gstack-free-test-${stamp}-${process.pid}-${logPathSequence}.log`);
+  return nextShardLogPath(directory, 'gstack-free-test');
 }
 
-function exitCodeFor(status: FreeShardStatus): number {
+export function exitCodeFor(status: FreeShardStatus): number {
   if (status === 'passed') return 0;
   return status === 'timed-out' ? 124 : 1;
 }
@@ -2072,16 +2043,8 @@ async function recordFreeTestDurations(files: string[], jobs: number): Promise<n
   }
 
   const target = process.env.GSTACK_FREE_TEST_DURATIONS ?? path.join(ROOT, FREE_TEST_DURATIONS_FILE);
-  const payload = {
-    version: 1,
-    recordedAt: new Date().toISOString(),
-    durations: Object.fromEntries(Object.entries(durations).sort(([a], [b]) => (a < b ? -1 : 1))),
-  };
-  // Atomic temp+rename (capture-context-budget's pattern): a killed recorder
-  // must never leave a truncated seed for loadFreeTestDurations to warn on.
-  const tmp = `${target}.tmp-${process.pid}`;
-  fs.writeFileSync(tmp, `${JSON.stringify(payload, null, 2)}\n`);
-  fs.renameSync(tmp, target);
+  // Atomic: a killed recorder never leaves a truncated seed behind.
+  writeDurationSeed(target, durations);
   console.log(`[test:free] wrote ${Object.keys(durations).length} durations to ${path.relative(ROOT, target)}`);
   if (failed.length > 0) {
     // Failures still recorded (a red file's duration is still a real cost),
diff --git a/scripts/test-paid-shards.ts b/scripts/test-paid-shards.ts
index cd1688c2e..eb80b1887 100644
--- a/scripts/test-paid-shards.ts
+++ b/scripts/test-paid-shards.ts
@@ -38,8 +38,9 @@
  *
  * Enumeration matches package.json's `test:gate` globs (via the shared
  * test/helpers/paid-test-set.ts) and honors EVALS_TIER against the E2E_TIERS
- * map in test/helpers/touchfiles.ts. Output classification reuses
- * scripts/test-strict-output.ts rather than reimplementing it.
+ * map in test/helpers/touchfiles.ts. Spawn, kill, sandbox, logs, seeds and
+ * output classification come from the shared shard engine
+ * (scripts/lib/shard-engine.ts); this file keeps only paid-lane policy.
  *
  * Parallelism now lives ACROSS shards (--jobs), not inside one Bun process, so
  * each shard runs its own file sequentially and can be killed independently.
@@ -57,14 +58,23 @@ import { spawnSync } from 'node:child_process';
 import { createBootstrapRetentionScope } from '../test/helpers/bootstrap-retention';
 import {
   BunTestOutputClassifier,
+  createShardSandbox,
   exactTestFileSelectors,
   forwardAndClassify,
   isTerminationRequested,
+  nextShardLogPath,
   normalizeRelativePath,
+  openShardLog,
+  parseCliFlags,
+  readDurationSeed,
+  removeShardSandbox,
   runShardChild,
-  strictTestExitCode,
+  strictShardStatus,
+  writeDurationSeed,
+  zeroExecutionVerdict,
+  type LanePolicy,
   type ShardChildResult,
-} from './test-strict-output';
+} from './lib/shard-engine';
 import { PAID_TEST_GLOBS, isPaidTestFile } from '../test/helpers/paid-test-set';
 import { PERIODIC_CI_EXCLUDE } from '../test/helpers/periodic-exclude-data';
 import { FILE_RETRY_BUDGETS, STRICT_RETRY_CASE_BUDGETS } from '../test/helpers/eval-budgets';
@@ -110,6 +120,17 @@ export const DEFAULT_MAX_FILES_PER_SHARD = 1;
 export const DEFAULT_JOBS = 8;
 export const DEFAULT_WITHIN_SHARD_CONCURRENCY = 2;
 
+/**
+ * Paid-lane classification policy. Seeds keep only positive walls (a zero
+ * is not a real paid-shard measurement). A shard that passed with zero
+ * executed tests is legitimate under selection (in-file diff/tier
+ * self-skips) and only warns; under EVALS_ALL it is hollow: 'passed-empty'.
+ */
+export const PAID_LANE_POLICY: LanePolicy = {
+  acceptsSeedDuration: (ms) => ms > 0,
+  zeroExecution: ({ promisedAll }) => (promisedAll ? 'passed-empty' : 'passed-with-warning'),
+};
+
 /** One overlay process preserves the original process-wide SDK semaphore. */
 export const OVERLAY_MAX_ACTIVE_SHARDS = 1;
 
@@ -654,15 +675,6 @@ export interface RunShardsOptions {
   casePatterns?: Record<string, string>;
 }
 
-let shardLogSequence = 0;
-
-/** Per-shard log path: slug + timestamp; pid + sequence defeat same-ms collisions. */
-function nextShardLogPath(files: string[], logDir: string): string {
-  const stamp = new Date().toISOString().replace(/[:.]/g, '-');
-  shardLogSequence += 1;
-  return path.join(logDir, `gstack-paid-shard-${shardSlug(files)}-${stamp}-${process.pid}-${shardLogSequence}.log`);
-}
-
 /** On-failure console excerpt budget: the last N bytes of the shard's log. */
 export const FAILURE_TAIL_BYTES = 64 * 1024;
 
@@ -712,15 +724,15 @@ export async function runPaidShard(
       ), ...(options.casePatterns ? ['--test-name-pattern', options.casePatterns[files[0]]] : [])],
     };
 
-  const env = { ...(options.env ?? process.env) };
+  const baseEnv = { ...(options.env ?? process.env) };
   if (options.evalDirBase) {
-    env.GSTACK_EVAL_DIR = path.join(options.evalDirBase, 'shards', shardSlug(files));
+    baseEnv.GSTACK_EVAL_DIR = path.join(options.evalDirBase, 'shards', shardSlug(files));
   }
   // Resolve `claude --version` ONCE in the parent (cached across shards) and
   // hand it to every child: eval-store's fallback is a synchronous spawn on
   // the same thread that polls PTY sessions, so children must never pay it.
-  if (!env.GSTACK_CLAUDE_CLI_VERSION) {
-    env.GSTACK_CLAUDE_CLI_VERSION = getClaudeCliVersion();
+  if (!baseEnv.GSTACK_CLAUDE_CLI_VERSION) {
+    baseEnv.GSTACK_CLAUDE_CLI_VERSION = getClaudeCliVersion();
   }
   // Per-shard temp + Chromium-profile isolation — the free runner treats
   // this as mandatory (test-free-shards.ts: two concurrent shards on one
@@ -731,13 +743,8 @@ export async function runPaidShard(
   // stopping wedged runs from accumulating full git-repo workspaces in the
   // shared tmpdir forever. Prerequisite for raising EVALS_JOBS (more
   // concurrency on shared state amplifies exactly the opus-47 race class).
-  const stateDir = fs.mkdtempSync(path.join(os.tmpdir(), 'gstack-paid-shard-'));
-  const childTmp = path.join(stateDir, 'tmp');
-  fs.mkdirSync(childTmp);
-  env.TMPDIR = childTmp;
-  env.TEMP = childTmp;
-  env.TMP = childTmp;
-  env.CHROMIUM_PROFILE = path.join(stateDir, 'chromium-profile');
+  const sandbox = createShardSandbox('gstack-paid-shard-', baseEnv);
+  const { stateDir, tmp: childTmp, env } = sandbox;
   const bootstrapFile = files.some(file => normalizeRelativePath(file) === 'test/skill-e2e-qa-workflow.test.ts');
   delete env.GSTACK_BOOTSTRAP_RETENTION;
   if (bootstrapFile && process.platform !== 'linux') log(`${label} bootstrap dependency retention unavailable on ${process.platform}; native behavior still runs without retained-dependency qualification`);
@@ -754,14 +761,9 @@ export async function runPaidShard(
   // model), never in a whole-run Buffer[] — non-live shards used to hold
   // their entire 30-min stream-json stdout+stderr in RAM, × concurrent jobs.
   // Printed at START so a wedged shard is inspectable live, mid-run.
-  const logPath = nextShardLogPath(files, options.logDir ?? os.tmpdir());
-  const logStream = fs.createWriteStream(logPath);
-  let logWriteFailed = false;
-  logStream.on('error', (err) => {
-    if (logWriteFailed) return;
-    logWriteFailed = true;
-    console.error(`${label} could not write the full log at ${logPath}: ${err.message}`);
-  });
+  const logPath = nextShardLogPath(options.logDir ?? os.tmpdir(), `gstack-paid-shard-${shardSlug(files)}`);
+  const spool = openShardLog(logPath, label);
+  const logStream = spool.stream;
   log(`${label} full log: ${logPath}`);
 
   const classifier = new BunTestOutputClassifier();
@@ -770,7 +772,7 @@ export async function runPaidShard(
   // verdict path is unchanged by where the bytes land afterwards.
   const sink = (destination: NodeJS.WriteStream): NodeJS.WriteStream => ({
     write: (chunk: Buffer | string): boolean => {
-      if (!logWriteFailed) logStream.write(chunk);
+      spool.write(chunk);
       if (streamLive) destination.write(chunk);
       return true;
     },
@@ -822,21 +824,21 @@ export async function runPaidShard(
         logStream.off('error', onError);
         logStream.off('close', onClose);
         if (!complete) {
-          logWriteFailed = true;
+          spool.failed = true;
           logStream.destroy();
           log(`${label} incomplete log capture; retained prefix: ${logPath}`);
         }
         resolve();
       };
       const onError = () => finish(false);
-      const onClose = () => finish(logStream.writableFinished && !logWriteFailed);
+      const onClose = () => finish(logStream.writableFinished && !spool.failed);
       const expire = () => { timedOut = true; finish(false); };
       logStream.once('error', onError);
       logStream.once('close', onClose);
       if (Date.now() >= shardDeadline) { expire(); return; }
-      if (logWriteFailed || logStream.destroyed) { finish(false); return; }
+      if (spool.failed || logStream.destroyed) { finish(false); return; }
       timer = setTimeout(expire, shardDeadline - Date.now());
-      try { logStream.end(() => finish(logStream.writableFinished && !logWriteFailed)); }
+      try { logStream.end(() => finish(logStream.writableFinished && !spool.failed)); }
       catch { finish(false); }
     });
     let retentionRemovable = true;
@@ -852,16 +854,8 @@ export async function runPaidShard(
         log(`${label} bootstrap retention acknowledgment failed; preserving shard state`);
       }
     }
-    try {
-      // async rm: a SIGKILLed shard can leave a full git workspace + Chromium
-      // profile here; a synchronous recursive delete on the parent's event
-      // loop would stall every sibling shard's stream classification and
-      // wall timers for seconds (review finding).
-      if (retentionRemovable) await fs.promises.rm(stateDir, { recursive: true, force: true });
-    } catch {
-      // Best-effort: a locked file must not turn a real verdict into an
-      // exception (same posture as the free runner's cleanup).
-    }
+    // Async, best-effort backstop: group-SIGKILLed tests never clean up.
+    if (retentionRemovable) await removeShardSandbox(stateDir);
   }
 
   const summary = classifier.end();
@@ -875,9 +869,10 @@ export async function runPaidShard(
   // commands must print a synthetic `Ran N tests across M files. [Xms]` line,
   // so tests can pin the summary-missing => failure backstop.
   const expectedFiles = files.length;
-  let status: ShardStatus = timedOut
-    ? 'timed-out'
-    : !retentionFailed && !logWriteFailed && !incompleteCapture && strictTestExitCode(exitCode ?? 1, summary, expectedFiles) === 0 ? 'passed' : 'failed';
+  let status: ShardStatus = strictShardStatus({
+    timedOut, exitCode, summary, expectedFiles,
+    evidenceComplete: !retentionFailed && !spool.failed && !incompleteCapture,
+  });
   if (status === 'passed' && options.expectedCases) {
     const expected = files.reduce((count, file) => count + (options.expectedCases![file] ?? 0), 0);
     const actual = summary.terminalTestCounts.reduce((count, value) => count + value, 0) - summary.skippedTests;
@@ -950,12 +945,13 @@ export function applyHollowShardGuard(
         (outcome.executedTests === null || outcome.executedTests === 0 || isAllSkippedPass(outcome))) {
       return { ...outcome, status: 'passed-empty' };
     }
-    if (outcome.status !== 'passed' || outcome.executedTests !== 0) return outcome;
-    if (!opts.evalsAll) {
+    if (outcome.status !== 'passed') return outcome;
+    const verdict = zeroExecutionVerdict(outcome.executedTests, PAID_LANE_POLICY, { promisedAll: opts.evalsAll });
+    if (verdict === 'passed-with-warning') {
       warn(`[test:paid] WARNING: shard ${outcome.shard} passed with 0 executed tests (${outcome.files.join(' ')}) — legitimate under selection, hollow under EVALS_ALL`);
       return outcome;
     }
-    return { ...outcome, status: 'passed-empty' };
+    return verdict === 'passed-empty' ? { ...outcome, status: 'passed-empty' } : outcome;
   });
 }
 
@@ -1114,13 +1110,8 @@ export const PAID_TEST_DURATIONS_FILE = 'scripts/paid-test-durations.json';
  * missing or corrupt seed keeps the supervision-budget allocation.
  */
 export function loadPaidTestDurations(rootDir = ROOT): Record<string, number> {
-  try {
-    const parsed = JSON.parse(fs.readFileSync(path.join(rootDir, PAID_TEST_DURATIONS_FILE), 'utf8')) as { durations?: Record<string, unknown> };
-    return Object.fromEntries(Object.entries(parsed.durations ?? {})
-      .filter((entry): entry is [string, number] => typeof entry[1] === 'number' && Number.isFinite(entry[1]) && entry[1] > 0));
-  } catch {
-    return {};
-  }
+  const seed = readDurationSeed(path.join(rootDir, PAID_TEST_DURATIONS_FILE), PAID_LANE_POLICY.acceptsSeedDuration);
+  return seed.status === 'ok' ? seed.durations : {};
 }
 
 /** Merge a report's executed single-file outcomes into the seed; all-skipped shards carry no cost signal. */
@@ -1556,44 +1547,32 @@ export function parseCliOptions(argv: string[], env: NodeJS.ProcessEnv = process
     writeDurations: false,
   };
 
-  for (let index = 0; index < argv.length; index += 1) {
-    const arg = argv[index];
-    if (arg === '--list') { options.listOnly = true; continue; }
-    if (arg === '--tier') {
-      const value = argv[index += 1];
+  const pathValue = (message: string, assign: (value: string) => void) => (next: () => string | undefined) => {
+    const value = next();
+    if (!value) throw new Error(message);
+    assign(value);
+  };
+  parseCliFlags(argv, {
+    '--list': () => { options.listOnly = true; },
+    '--tier': (next) => {
+      const value = next();
       if (value !== 'gate' && value !== 'periodic') throw new Error(`--tier must be gate or periodic. Received: ${value}`);
       options.tier = value;
-      continue;
-    }
-    if (arg === '--profile') {
-      const value = argv[index += 1];
-      if (!value) throw new Error('--profile needs pr or full');
-      options.profile = validatedProfile(value, '--profile'); options.profileExplicit = true; continue;
-    }
-    if (arg === '--timeout') { options.timeoutMs = parsePositiveInt(argv[index += 1], '--timeout') * 1000; options.timeoutExplicit = true; continue; }
-    if (arg === '--jobs') { options.jobs = parsePositiveInt(argv[index += 1], '--jobs'); continue; }
-    if (arg === '--files-per-shard') { options.maxFilesPerShard = parsePositiveInt(argv[index += 1], '--files-per-shard'); continue; }
-    if (arg === '--emit-plan') {
-      const value = argv[index += 1];
-      if (!value) throw new Error('--emit-plan needs a file path');
-      options.emitPlanPath = value; continue;
-    }
-    if (arg === '--skip-judges') { options.skipJudges = true; continue; }
-    if (arg === '--slices') { options.slices = parsePositiveInt(argv[index += 1], '--slices'); continue; }
-    if (arg === '--plan') {
-      const value = argv[index += 1];
-      if (!value) throw new Error('--plan needs a manifest path');
-      options.planPath = value; continue;
-    }
-    if (arg === '--slice') { options.sliceIndex = parsePositiveInt(argv[index += 1], '--slice'); continue; }
-    if (arg === '--report') {
-      const value = argv[index += 1];
-      if (!value) throw new Error('--report needs a directory');
-      options.reportDir = value; continue;
-    }
-    if (arg === '--write-durations') { options.writeDurations = true; continue; }
-    throw new Error(`Unknown argument: ${arg}`);
-  }
+    },
+    '--profile': pathValue('--profile needs pr or full', (value) => {
+      options.profile = validatedProfile(value, '--profile'); options.profileExplicit = true;
+    }),
+    '--timeout': (next) => { options.timeoutMs = parsePositiveInt(next(), '--timeout') * 1000; options.timeoutExplicit = true; },
+    '--jobs': (next) => { options.jobs = parsePositiveInt(next(), '--jobs'); },
+    '--files-per-shard': (next) => { options.maxFilesPerShard = parsePositiveInt(next(), '--files-per-shard'); },
+    '--emit-plan': pathValue('--emit-plan needs a file path', (value) => { options.emitPlanPath = value; }),
+    '--skip-judges': () => { options.skipJudges = true; },
+    '--slices': (next) => { options.slices = parsePositiveInt(next(), '--slices'); },
+    '--plan': pathValue('--plan needs a manifest path', (value) => { options.planPath = value; }),
+    '--slice': (next) => { options.sliceIndex = parsePositiveInt(next(), '--slice'); },
+    '--report': pathValue('--report needs a directory', (value) => { options.reportDir = value; }),
+    '--write-durations': () => { options.writeDurations = true; },
+  });
   if (options.writeDurations && !options.reportDir) throw new Error('--write-durations requires --report');
   if (options.skipJudges && (!options.emitPlanPath || options.tier !== 'gate')) throw new Error('--skip-judges applies only to an emitted gate census plan');
   if (options.profile === 'pr' && options.tier !== 'gate') throw new Error('PR profile requires gate tier');
@@ -1648,10 +1627,7 @@ async function main(): Promise<number> {
     }
     if (options.writeDurations) {
       const durations = mergePaidTestDurations(loadPaidTestDurations(), results);
-      const target = path.join(ROOT, PAID_TEST_DURATIONS_FILE);
-      const temporary = `${target}.tmp-${process.pid}`;
-      fs.writeFileSync(temporary, `${JSON.stringify({ version: 1, recordedAt: new Date().toISOString(), durations }, null, 2)}\n`);
-      fs.renameSync(temporary, target);
+      writeDurationSeed(path.join(ROOT, PAID_TEST_DURATIONS_FILE), durations);
       console.log(`[test:paid] wrote ${Object.keys(durations).length} durations to ${PAID_TEST_DURATIONS_FILE}`);
     }
     // Historical flaky_retries includes every case with multiple attempts,
diff --git a/test/paid-free-boundary.test.ts b/test/paid-free-boundary.test.ts
index 71a5350d3..3186a23f1 100644
--- a/test/paid-free-boundary.test.ts
+++ b/test/paid-free-boundary.test.ts
@@ -64,7 +64,7 @@ describe('paid/free dependency boundary', () => {
     expect(paid.length).toBeGreaterThan(0);
     const all = workflowJudgeDependencies(ROOT, paid);
     for (const dependencies of [runner, all]) {
-      expect(dependencies).toContain('scripts/test-strict-output.ts');
+      expect(dependencies).toContain('scripts/lib/shard-engine.ts');
       expect(dependencies).toContain('test/helpers/test-selection.ts');
       expect(dependencies).not.toContain('scripts/eval-flake-rank.ts');
       for (const freeOnly of FREE_ONLY_PR_FILES) expect(dependencies).not.toContain(freeOnly);