@moretea-labs/forge 1.7.2 → 1.8.1

package/scripts/benchmark-ios-agent-device.ts

@@ -1,514 +0,0 @@
-#!/usr/bin/env bun
-/**
- * Real iOS agent-device latency benchmark.
- *
- * Compares the compatibility fresh-session JD workflow with the trusted warm
- * path that reuses one exact active JD interaction. It never invents baseline
- * data and never runs without an explicitly selected --device. Every selected
- * profile performs one warmup followed by at least three measured runs.
- *
- * Fresh-session profile:
- *   inventory -> open -> search/evidence -> screenshot -> close, for every run
- *
- * Reused-session profile:
- *   open once -> repeated search/evidence batches without screenshot -> close once
- *
- * Example:
- *   bun scripts/benchmark-ios-agent-device.ts \
- *     --repo-id repo_... --device "iPhone" --query "宽楦男鞋" \
- *     --search-selector 'type="SearchField"' --submit-selector 'label="搜索"' \
- *     --result-text "宽楦男鞋" --profile both --runs 5 --json
- */
-import { performance } from 'perf_hooks';
-import { resolve } from 'path';
-import { executeIosAgentDeviceAction } from '../src/runtime/plugins/ios-agent-device';
-
-type EvidenceTier = 'exact_wait' | 'scoped_snapshot' | 'full_snapshot' | 'unknown';
-export type BenchmarkProfile = 'fresh_session' | 'reused_session';
-
-export interface NumericSummary {
-  count: number;
-  min: number;
-  p50: number;
-  p95: number;
-  max: number;
-  mean: number;
-}
-
-interface PhaseTimings {
-  targetSelection?: number;
-  open?: number;
-  targetDiscovery?: number;
-  interactionAndEvidence?: number;
-  screenshot?: number;
-  close?: number;
-  total?: number;
-}
-
-export interface IosBenchmarkSample {
-  index: number;
-  warmup: boolean;
-  profile: BenchmarkProfile;
-  totalMs: number;
-  harnessOverheadMs?: number;
-  tier: EvidenceTier;
-  sessionReused: boolean;
-  sessionKept: boolean;
-  screenshotCaptured: boolean;
-  deviceInventoryRequests: number;
-  accessibilitySnapshotRequests: number;
-  nativeBatchRequests: number;
-  nativeBatchSteps: number;
-  staleRefRecovery: boolean;
-  exactWaitFallback: boolean;
-  runnerRoundTrips?: number;
-  providerWallClockMs?: number;
-  phaseTimingsMs?: PhaseTimings;
-}
-
-export interface IosBenchmarkProfileReport {
-  profile: BenchmarkProfile;
-  warmupCount: number;
-  measuredRunCount: number;
-  setup?: { openMs: number };
-  cleanup?: { closeMs: number; completed: boolean };
-  measured: Record<string, unknown>;
-  samples: IosBenchmarkSample[];
-}
-
-const JD_BUNDLE_ID = 'com.360buy.jdmobile';
-
-function rounded(value: number): number {
-  return Math.round(value * 100) / 100;
-}
-
-export function percentile(values: number[], percentileValue: number): number {
-  if (values.length === 0) return 0;
-  const sorted = [...values].sort((left, right) => left - right);
-  const rank = Math.max(0, Math.min(1, percentileValue)) * (sorted.length - 1);
-  const lower = Math.floor(rank);
-  const upper = Math.ceil(rank);
-  if (lower === upper) return sorted[lower]!;
-  const weight = rank - lower;
-  return sorted[lower]! * (1 - weight) + sorted[upper]! * weight;
-}
-
-export function summarize(values: number[]): NumericSummary {
-  if (values.length === 0) return { count: 0, min: 0, p50: 0, p95: 0, max: 0, mean: 0 };
-  return {
-    count: values.length,
-    min: rounded(Math.min(...values)),
-    p50: rounded(percentile(values, 0.5)),
-    p95: rounded(percentile(values, 0.95)),
-    max: rounded(Math.max(...values)),
-    mean: rounded(values.reduce((sum, value) => sum + value, 0) / values.length),
-  };
-}
-
-function option(name: string): string | undefined {
-  const index = process.argv.indexOf(`--${name}`);
-  return index >= 0 ? process.argv[index + 1] : undefined;
-}
-
-function hasFlag(name: string): boolean {
-  return process.argv.includes(`--${name}`);
-}
-
-function integerOption(name: string, fallback: number, minimum: number, maximum: number): number {
-  const raw = option(name);
-  if (raw === undefined) return fallback;
-  const value = Number(raw);
-  if (!Number.isInteger(value) || value < minimum || value > maximum) {
-    throw new Error(`--${name} must be an integer between ${minimum} and ${maximum}`);
-  }
-  return value;
-}
-
-function requiredOption(name: string): string {
-  const value = option(name)?.trim();
-  if (!value) throw new Error(`--${name} is required`);
-  return value;
-}
-
-function selectedProfiles(): BenchmarkProfile[] {
-  const value = (option('profile') ?? 'both').trim().toLowerCase();
-  if (value === 'both') return ['fresh_session', 'reused_session'];
-  if (value === 'fresh' || value === 'fresh_session') return ['fresh_session'];
-  if (value === 'warm' || value === 'reused' || value === 'reused_session') return ['reused_session'];
-  throw new Error('--profile must be fresh, warm, or both');
-}
-
-function stringRecord(value: unknown): Record<string, unknown> | undefined {
-  return value && typeof value === 'object' && !Array.isArray(value)
-    ? value as Record<string, unknown>
-    : undefined;
-}
-
-function findCost(value: unknown): { wallClockMs?: number; runnerRoundTrips?: number } {
-  if (Array.isArray(value)) {
-    for (const item of value) {
-      const found = findCost(item);
-      if (found.wallClockMs !== undefined || found.runnerRoundTrips !== undefined) return found;
-    }
-    return {};
-  }
-  const record = stringRecord(value);
-  if (!record) return {};
-  const cost = stringRecord(record.cost);
-  if (cost) {
-    return {
-      wallClockMs: typeof cost.wallClockMs === 'number' ? cost.wallClockMs : undefined,
-      runnerRoundTrips: typeof cost.runnerRoundTrips === 'number' ? cost.runnerRoundTrips : undefined,
-    };
-  }
-  for (const child of Object.values(record)) {
-    const found = findCost(child);
-    if (found.wallClockMs !== undefined || found.runnerRoundTrips !== undefined) return found;
-  }
-  return {};
-}
-
-function sampleFromResult(
-  profile: BenchmarkProfile,
-  index: number,
-  warmup: boolean,
-  totalMs: number,
-  result: Record<string, unknown>,
-): IosBenchmarkSample {
-  const executionPlan = stringRecord(result.executionPlan) ?? {};
-  const phaseTimingsMs = stringRecord(executionPlan.timingsMs) as PhaseTimings | undefined;
-  const providerTotal = phaseTimingsMs?.total;
-  const cost = findCost(result);
-  return {
-    index,
-    warmup,
-    profile,
-    totalMs: rounded(totalMs),
-    harnessOverheadMs: typeof providerTotal === 'number'
-      ? rounded(Math.max(0, totalMs - providerTotal))
-      : undefined,
-    tier: typeof executionPlan.accessibilityEvidenceTier === 'string'
-      ? executionPlan.accessibilityEvidenceTier as EvidenceTier
-      : 'unknown',
-    sessionReused: executionPlan.sessionReused === true,
-    sessionKept: executionPlan.sessionKept === true,
-    screenshotCaptured: executionPlan.screenshotCaptured === true,
-    deviceInventoryRequests: Number(executionPlan.deviceInventoryRequests ?? 0),
-    accessibilitySnapshotRequests: Number(executionPlan.accessibilitySnapshotRequests ?? 0),
-    nativeBatchRequests: Number(executionPlan.nativeBatchRequests ?? 0),
-    nativeBatchSteps: Number(executionPlan.nativeBatchSteps ?? 0),
-    staleRefRecovery: executionPlan.staleRefRecovery === true,
-    exactWaitFallback: executionPlan.exactWaitFallback === true,
-    runnerRoundTrips: cost.runnerRoundTrips,
-    providerWallClockMs: cost.wallClockMs,
-    phaseTimingsMs,
-  };
-}
-
-const PHASE_NAMES: Array<keyof PhaseTimings> = [
-  'targetSelection',
-  'open',
-  'targetDiscovery',
-  'interactionAndEvidence',
-  'screenshot',
-  'close',
-  'total',
-];
-
-export function summarizeProfileSamples(samples: IosBenchmarkSample[]): Record<string, unknown> {
-  const measured = samples.filter((sample) => !sample.warmup);
-  const phaseTimingsMs = Object.fromEntries(PHASE_NAMES.map((phase) => [
-    phase,
-    summarize(measured.flatMap((sample) => {
-      const value = sample.phaseTimingsMs?.[phase];
-      return typeof value === 'number' ? [value] : [];
-    })),
-  ]));
-  const byTier = Object.fromEntries(
-    [...new Set(measured.map((sample) => sample.tier))].map((tier) => [
-      tier,
-      {
-        totalMs: summarize(measured.filter((sample) => sample.tier === tier).map((sample) => sample.totalMs)),
-        providerWallClockMs: summarize(measured
-          .filter((sample) => sample.tier === tier && sample.providerWallClockMs !== undefined)
-          .map((sample) => sample.providerWallClockMs!)),
-        runnerRoundTrips: summarize(measured
-          .filter((sample) => sample.tier === tier && sample.runnerRoundTrips !== undefined)
-          .map((sample) => sample.runnerRoundTrips!)),
-      },
-    ]),
-  );
-  return {
-    totalMs: summarize(measured.map((sample) => sample.totalMs)),
-    harnessOverheadMs: summarize(measured
-      .filter((sample) => sample.harnessOverheadMs !== undefined)
-      .map((sample) => sample.harnessOverheadMs!)),
-    deviceInventoryRequests: summarize(measured.map((sample) => sample.deviceInventoryRequests)),
-    accessibilitySnapshotRequests: summarize(measured.map((sample) => sample.accessibilitySnapshotRequests)),
-    nativeBatchRequests: summarize(measured.map((sample) => sample.nativeBatchRequests)),
-    nativeBatchSteps: summarize(measured.map((sample) => sample.nativeBatchSteps)),
-    runnerRoundTrips: summarize(measured
-      .filter((sample) => sample.runnerRoundTrips !== undefined)
-      .map((sample) => sample.runnerRoundTrips!)),
-    phaseTimingsMs,
-    byTier,
-    invariants: {
-      allSessionsReused: measured.length > 0 && measured.every((sample) => sample.sessionReused),
-      allScreenshotsSkipped: measured.length > 0 && measured.every((sample) => !sample.screenshotCaptured),
-      allInventorySkipped: measured.length > 0 && measured.every((sample) => sample.deviceInventoryRequests === 0),
-      maxNativeBatchRequests: measured.length > 0 ? Math.max(...measured.map((sample) => sample.nativeBatchRequests)) : 0,
-    },
-  };
-}
-
-export function compareProfileReports(
-  fresh: IosBenchmarkProfileReport | undefined,
-  reused: IosBenchmarkProfileReport | undefined,
-): Record<string, unknown> | undefined {
-  if (!fresh || !reused) return undefined;
-  const freshTotal = (fresh.measured.totalMs as NumericSummary).p50;
-  const reusedTotal = (reused.measured.totalMs as NumericSummary).p50;
-  const freshP95 = (fresh.measured.totalMs as NumericSummary).p95;
-  const reusedP95 = (reused.measured.totalMs as NumericSummary).p95;
-  return {
-    p50SavedMs: rounded(freshTotal - reusedTotal),
-    p95SavedMs: rounded(freshP95 - reusedP95),
-    p50Speedup: reusedTotal > 0 ? rounded(freshTotal / reusedTotal) : null,
-    p95Speedup: reusedP95 > 0 ? rounded(freshP95 / reusedP95) : null,
-    note: 'Fresh-session includes per-run inventory/open/screenshot/close; reused-session excludes one-time setup and cleanup from measured search latency.',
-  };
-}
-
-function usage(): string {
-  return [
-    'Usage: bun scripts/benchmark-ios-agent-device.ts --repo-id <id> --device <name> --query <text> [options]',
-    '',
-    'Required:',
-    '  --repo-id <id>             Repository id used for Controller-owned interaction state',
-    '  --device <name>            Exact connected physical iPhone or booted Simulator name',
-    '  --query <text>              Non-sensitive product-information query',
-    '',
-    'Evidence targeting:',
-    '  --search-selector <value>   Stable selector; preferred over snapshot-scoped refs',
-    '  --search-target <ref>       Cached accessibility ref fallback',
-    '  --submit-selector <value>   Stable submit selector',
-    '  --submit-target <ref>       Cached submit ref fallback',
-    '  --result-text <text>        Exact text wait',
-    '  --result-selector <value>   Exact selector wait',
-    '  --result-scope <value>      Scoped snapshot fallback',
-    '  --snapshot-depth <1..20>    Snapshot depth, default 8',
-    '',
-    'Execution:',
-    '  --profile <fresh|warm|both> Profiles to measure, default both',
-    '  --runs <3..20>              Measured runs per profile after one warmup, default 3',
-    '  --timeout-ms <ms>           Absolute timeout per action, default 60000',
-    '  --controller-home <path>    Defaults to FORGE_CONTROLLER_HOME or _ops/controller-home',
-    '  --repo-root <path>          Defaults to current working directory',
-    '  --relaunch                  Relaunch JD for each fresh run and the warm-profile setup open',
-    '  --json                      Emit machine-readable JSON',
-  ].join('\n');
-}
-
-interface BenchmarkContext {
-  controllerHome: string;
-  repoId: string;
-  repoRoot: string;
-  device: string;
-  query: string;
-  timeoutMs: number;
-  relaunch: boolean;
-  runs: number;
-  baseArgs: Record<string, unknown>;
-  runNonce: string;
-}
-
-async function executeAction(
-  context: BenchmarkContext,
-  actionId: string,
-  suffix: string,
-  args: Record<string, unknown>,
-): Promise<Record<string, unknown>> {
-  return executeIosAgentDeviceAction({
-    controllerHome: context.controllerHome,
-    repoId: context.repoId,
-    repoRoot: context.repoRoot,
-    pluginId: 'ios',
-    actionId,
-    requestId: `ios-latency-benchmark-${context.runNonce}-${suffix}`,
-    args,
-    timeoutMs: context.timeoutMs,
-    origin: { surface: 'local-ui', actor: 'ios-latency-benchmark' },
-  });
-}
-
-async function executeSearchSample(
-  context: BenchmarkContext,
-  profile: BenchmarkProfile,
-  index: number,
-  args: Record<string, unknown>,
-): Promise<IosBenchmarkSample> {
-  const startedAt = performance.now();
-  const result = await executeAction(context, 'agent_device_jd_search', `${profile}-${index}`, args);
-  return sampleFromResult(profile, index, index === 0, performance.now() - startedAt, result);
-}
-
-async function runFreshSessionProfile(context: BenchmarkContext): Promise<IosBenchmarkProfileReport> {
-  const samples: IosBenchmarkSample[] = [];
-  for (let index = 0; index <= context.runs; index += 1) {
-    samples.push(await executeSearchSample(context, 'fresh_session', index, {
-      ...context.baseArgs,
-      relaunch: context.relaunch,
-      keep_session: false,
-      capture_screenshot: true,
-    }));
-  }
-  return {
-    profile: 'fresh_session',
-    warmupCount: 1,
-    measuredRunCount: context.runs,
-    measured: summarizeProfileSamples(samples),
-    samples,
-  };
-}
-
-async function runReusedSessionProfile(context: BenchmarkContext): Promise<IosBenchmarkProfileReport> {
-  const openStartedAt = performance.now();
-  const opened = await executeAction(context, 'agent_device_open', 'warm-open', {
-    app: JD_BUNDLE_ID,
-    device: context.device,
-    relaunch: context.relaunch,
-  });
-  const openMs = rounded(performance.now() - openStartedAt);
-  const interaction = stringRecord(opened.interaction);
-  const interactionId = typeof interaction?.interactionId === 'string' ? interaction.interactionId : undefined;
-  if (!interactionId) throw new Error('agent_device_open did not return an interaction_id');
-
-  const samples: IosBenchmarkSample[] = [];
-  let closeMs = 0;
-  let cleanupCompleted = false;
-  try {
-    for (let index = 0; index <= context.runs; index += 1) {
-      samples.push(await executeSearchSample(context, 'reused_session', index, {
-        ...context.baseArgs,
-        interaction_id: interactionId,
-        keep_session: true,
-        capture_screenshot: false,
-        relaunch: false,
-      }));
-    }
-  } finally {
-    const closeStartedAt = performance.now();
-    try {
-      await executeAction(context, 'agent_device_close', 'warm-close', { interaction_id: interactionId });
-      cleanupCompleted = true;
-    } finally {
-      closeMs = rounded(performance.now() - closeStartedAt);
-    }
-  }
-  return {
-    profile: 'reused_session',
-    warmupCount: 1,
-    measuredRunCount: context.runs,
-    setup: { openMs },
-    cleanup: { closeMs, completed: cleanupCompleted },
-    measured: summarizeProfileSamples(samples),
-    samples,
-  };
-}
-
-export async function runBenchmark(): Promise<Record<string, unknown>> {
-  if (hasFlag('help')) {
-    console.log(usage());
-    return { help: true };
-  }
-  const repoId = requiredOption('repo-id');
-  const device = requiredOption('device');
-  const query = requiredOption('query');
-  const runs = integerOption('runs', 3, 3, 20);
-  const timeoutMs = integerOption('timeout-ms', 60_000, 1_000, 600_000);
-  const snapshotDepth = integerOption('snapshot-depth', 8, 1, 20);
-  const repoRoot = resolve(option('repo-root') ?? process.cwd());
-  const controllerHome = resolve(
-    option('controller-home')
-      ?? process.env.FORGE_CONTROLLER_HOME
-      ?? resolve(repoRoot, '_ops/controller-home'),
-  );
-  const baseArgs: Record<string, unknown> = { device, query, snapshot_depth: snapshotDepth };
-  for (const [flag, key] of [
-    ['search-selector', 'search_selector'],
-    ['search-target', 'search_target'],
-    ['submit-selector', 'submit_selector'],
-    ['submit-target', 'submit_target'],
-    ['result-text', 'result_text'],
-    ['result-selector', 'result_selector'],
-    ['result-scope', 'result_scope'],
-  ] as const) {
-    const value = option(flag)?.trim();
-    if (value) baseArgs[key] = value;
-  }
-  const profiles = selectedProfiles();
-  const context: BenchmarkContext = {
-    controllerHome,
-    repoId,
-    repoRoot,
-    device,
-    query,
-    timeoutMs,
-    relaunch: hasFlag('relaunch'),
-    runs,
-    baseArgs,
-    runNonce: `${Date.now()}-${process.pid}`,
-  };
-
-  const reports: Partial<Record<BenchmarkProfile, IosBenchmarkProfileReport>> = {};
-  for (const profile of profiles) {
-    reports[profile] = profile === 'fresh_session'
-      ? await runFreshSessionProfile(context)
-      : await runReusedSessionProfile(context);
-  }
-  const comparison = compareProfileReports(reports.fresh_session, reports.reused_session);
-  return {
-    schemaVersion: 2,
-    kind: 'ios_agent_device_latency_benchmark',
-    generatedAt: new Date().toISOString(),
-    environment: {
-      repoId,
-      device,
-      relaunch: context.relaunch,
-      profiles,
-      measuredRunsPerProfile: runs,
-      warmupsPerProfile: 1,
-      timeoutMs,
-      evidenceInputs: {
-        hasSearchSelector: typeof baseArgs.search_selector === 'string',
-        hasSearchTarget: typeof baseArgs.search_target === 'string',
-        hasSubmitSelector: typeof baseArgs.submit_selector === 'string',
-        hasSubmitTarget: typeof baseArgs.submit_target === 'string',
-        hasResultText: typeof baseArgs.result_text === 'string',
-        hasResultSelector: typeof baseArgs.result_selector === 'string',
-        resultScope: baseArgs.result_scope ?? null,
-        snapshotDepth,
-      },
-    },
-    profiles: reports,
-    ...(comparison ? { comparison } : {}),
-    note: 'Results are actual provider executions. No historical or unavailable device data is synthesized. Fresh-session and reused-session have intentionally different setup, screenshot, and cleanup costs.',
-  };
-}
-
-if (import.meta.main) {
-  try {
-    const report = await runBenchmark();
-    if (!('help' in report)) {
-      if (hasFlag('json')) console.log(JSON.stringify(report, null, 2));
-      else {
-        console.log('iOS agent-device latency benchmark');
-        console.log(JSON.stringify({ profiles: report.profiles, comparison: report.comparison }, null, 2));
-      }
-    }
-  } catch (error) {
-    console.error(error instanceof Error ? error.message : String(error));
-    console.error('\n' + usage());
-    process.exitCode = 1;
-  }
-}

package/scripts/benchmark-mcp-host.sh

@@ -1,117 +0,0 @@
-#!/usr/bin/env bash
-set -euo pipefail
-
-ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
-BUN_BIN="${BUN_BIN:-bun}"
-LABEL="${FORGE_MCP_BENCH_LABEL:-host-runtime-direct}"
-PORT="${FORGE_MCP_BENCH_PORT:-18765}"
-ITERATIONS="${FORGE_MCP_BENCH_ITERATIONS:-50}"
-CONNECT_ITERATIONS="${FORGE_MCP_BENCH_CONNECT_ITERATIONS:-30}"
-
-for value in "$PORT" "$ITERATIONS" "$CONNECT_ITERATIONS"; do
-  if ! [[ "$value" =~ ^[0-9]+$ ]] || [[ "$value" == "0" ]]; then
-    echo "benchmark:mcp-host requires positive integer port/iteration values" >&2
-    exit 2
-  fi
-done
-if (( PORT < 1024 || PORT > 65535 )); then
-  echo "FORGE_MCP_BENCH_PORT must be between 1024 and 65535" >&2
-  exit 2
-fi
-
-TMP_ROOT="$(mktemp -d "${TMPDIR:-/tmp}/forge-mcp-host-bench.XXXXXX")"
-CONTROLLER_HOME="$TMP_ROOT/controller-home"
-REGISTER_JSON="$TMP_ROOT/repository.json"
-INSTALL_JSON="$TMP_ROOT/runtime-install.json"
-READY_JSON="$TMP_ROOT/runtime-ready.json"
-STEADY_JSON="$TMP_ROOT/steady.json"
-CONNECT_JSON="$TMP_ROOT/connect.json"
-RUNTIME_PID=""
-
-cleanup() {
-  if [[ -n "$RUNTIME_PID" ]] && kill -0 "$RUNTIME_PID" 2>/dev/null; then
-    kill "$RUNTIME_PID" 2>/dev/null || true
-    for _ in $(seq 1 20); do
-      kill -0 "$RUNTIME_PID" 2>/dev/null || break
-      sleep 0.1
-    done
-    kill -9 "$RUNTIME_PID" 2>/dev/null || true
-  fi
-  for _ in $(seq 1 5); do
-    rm -rf "$TMP_ROOT" 2>/dev/null && return 0
-    sleep 0.1
-  done
-  rm -rf "$TMP_ROOT" 2>/dev/null || true
-}
-trap cleanup EXIT INT TERM
-
-TOKEN="$(node -e "process.stdout.write(require('crypto').randomBytes(32).toString('hex'))")"
-export FORGE_MCP_TOKEN="$TOKEN"
-export FORGE_MCP_BENCH_TOKEN="$TOKEN"
-
-cd "$ROOT"
-"$BUN_BIN" bin/forge.mjs repo register "$ROOT" \
-  --controller-home "$CONTROLLER_HOME" \
-  --json > "$REGISTER_JSON"
-REPO_ID="$(node -e "const x=require(process.argv[1]); const id=x?.repository?.repoId; if(!id) process.exit(2); process.stdout.write(id)" "$REGISTER_JSON")"
-
-"$BUN_BIN" bin/forge.mjs runtime service install-package \
-  --controller-home "$CONTROLLER_HOME" \
-  --host 127.0.0.1 \
-  --port "$PORT" \
-  --portable > "$INSTALL_JSON"
-RUNTIME_PID="$(node -e "const x=require(process.argv[1]); const pid=x?.pid; if(!Number.isInteger(pid)||pid<=0) process.exit(2); process.stdout.write(String(pid))" "$INSTALL_JSON")"
-
-curl --fail --silent --show-error \
-  --retry 30 --retry-delay 1 --retry-connrefused \
-  "http://127.0.0.1:${PORT}/ready" > "$READY_JSON"
-
-"$BUN_BIN" run benchmark:mcp-transport -- \
-  --endpoint "http://127.0.0.1:${PORT}/mcp" \
-  --label "$LABEL" \
-  --repo-id "$REPO_ID" \
-  --tool rh_status \
-  --iterations "$ITERATIONS" \
-  --warmup 3 \
-  --timing-log "$CONTROLLER_HOME/audit/mcp-timings.jsonl" > "$STEADY_JSON"
-
-"$BUN_BIN" run benchmark:mcp-transport -- \
-  --endpoint "http://127.0.0.1:${PORT}/mcp" \
-  --label "$LABEL" \
-  --repo-id "$REPO_ID" \
-  --tool rh_status \
-  --iterations "$CONNECT_ITERATIONS" \
-  --warmup 0 \
-  --include-connect \
-  --timing-log "$CONTROLLER_HOME/audit/mcp-timings.jsonl" > "$CONNECT_JSON"
-
-node - "$STEADY_JSON" "$CONNECT_JSON" "$READY_JSON" <<'NODE'
-const fs = require('fs');
-const os = require('os');
-const path = require('path');
-const [steadyPath, connectPath, readyPath] = process.argv.slice(2);
-const steady = JSON.parse(fs.readFileSync(steadyPath, 'utf8'));
-const connect = JSON.parse(fs.readFileSync(connectPath, 'utf8'));
-const ready = JSON.parse(fs.readFileSync(readyPath, 'utf8'));
-const report = {
-  schemaVersion: 1,
-  generatedAt: new Date().toISOString(),
-  host: {
-    platform: process.platform,
-    arch: process.arch,
-    cpuCount: os.cpus().length,
-    totalMemoryBytes: os.totalmem(),
-  },
-  ready,
-  steady,
-  connectInclusive: connect,
-};
-const text = `${JSON.stringify(report, null, 2)}\n`;
-const output = process.env.FORGE_MCP_BENCH_OUTPUT?.trim();
-if (output) {
-  const target = path.resolve(output);
-  fs.mkdirSync(path.dirname(target), { recursive: true });
-  fs.writeFileSync(target, text, 'utf8');
-}
-process.stdout.write(text);
-NODE