clawbench-cli 0.1.2__py3-none-any.whl

clawbench/data/chrome-extension/setup.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+set -e
+
+EXT_DIR="$(cd "$(dirname "$0")" && pwd)"
+
+# Detect Chrome binary
+if [[ "$OSTYPE" == "darwin"* ]]; then
+  CHROME="/Applications/Google Chrome.app/Contents/MacOS/Google Chrome"
+elif command -v google-chrome-stable &>/dev/null; then
+  CHROME="google-chrome-stable"
+elif command -v google-chrome &>/dev/null; then
+  CHROME="google-chrome"
+elif command -v chromium-browser &>/dev/null; then
+  CHROME="chromium-browser"
+elif command -v chromium &>/dev/null; then
+  CHROME="chromium"
+else
+  echo "Chrome not found" && exit 1
+fi
+
+"$CHROME" \
+  --no-first-run \
+  --disable-default-apps \
+  --remote-debugging-port=9222 \
+  --load-extension="$EXT_DIR" \
+  --disable-extensions-except="$EXT_DIR" \
+  "$@"

clawbench/data/chrome-extension/stealth.js

@@ -0,0 +1,200 @@
+/**
+ * stealth.js — Anti-bot-detection patches.
+ *
+ * Injected at document_start in the MAIN world (see manifest.json).
+ * Must run before any page script to reliably override browser fingerprints.
+ */
+
+(function () {
+  "use strict";
+
+  // 1. navigator.webdriver → false (real Chrome returns false, not undefined)
+  try {
+    Object.defineProperty(navigator, "webdriver", {
+      get: () => false,
+      configurable: true,
+    });
+  } catch (_) {}
+
+  // 2. navigator.languages — ensure realistic value
+  try {
+    Object.defineProperty(navigator, "languages", {
+      get: () => ["en-US", "en"],
+      configurable: true,
+    });
+  } catch (_) {}
+
+  // 3. navigator.plugins — fake standard Chrome plugins
+  try {
+    const fakePlugins = [
+      { name: "Chrome PDF Plugin", filename: "internal-pdf-viewer", description: "Portable Document Format" },
+      { name: "Chrome PDF Viewer", filename: "mhjfbmdgcfjbbpaeojofohoefgiehjai", description: "" },
+      { name: "Native Client", filename: "internal-nacl-plugin", description: "" },
+    ];
+
+    const pluginArray = Object.create(PluginArray.prototype);
+    fakePlugins.forEach((p, i) => {
+      const plugin = Object.create(Plugin.prototype);
+      Object.defineProperties(plugin, {
+        name: { get: () => p.name },
+        filename: { get: () => p.filename },
+        description: { get: () => p.description },
+        length: { get: () => 0 },
+      });
+      Object.defineProperty(pluginArray, i, { get: () => plugin, enumerable: true });
+    });
+    Object.defineProperty(pluginArray, "length", { get: () => fakePlugins.length });
+    pluginArray.item = (i) => pluginArray[i] || null;
+    pluginArray.namedItem = (name) => {
+      const idx = fakePlugins.findIndex((p) => p.name === name);
+      return idx >= 0 ? pluginArray[idx] : null;
+    };
+    pluginArray.refresh = () => {};
+    pluginArray[Symbol.iterator] = function* () {
+      for (let i = 0; i < fakePlugins.length; i++) yield pluginArray[i];
+    };
+
+    Object.defineProperty(navigator, "plugins", {
+      get: () => pluginArray,
+      configurable: true,
+    });
+  } catch (_) {}
+
+  // 4. navigator.mimeTypes — match the fake plugins
+  try {
+    const fakeMimeTypes = [
+      { type: "application/pdf", suffixes: "pdf", description: "Portable Document Format" },
+      { type: "application/x-google-chrome-pdf", suffixes: "pdf", description: "Portable Document Format" },
+    ];
+
+    const mimeTypeArray = Object.create(MimeTypeArray.prototype);
+    fakeMimeTypes.forEach((m, i) => {
+      const mimeType = Object.create(MimeType.prototype);
+      Object.defineProperties(mimeType, {
+        type: { get: () => m.type },
+        suffixes: { get: () => m.suffixes },
+        description: { get: () => m.description },
+        enabledPlugin: { get: () => null },
+      });
+      Object.defineProperty(mimeTypeArray, i, { get: () => mimeType, enumerable: true });
+    });
+    Object.defineProperty(mimeTypeArray, "length", { get: () => fakeMimeTypes.length });
+    mimeTypeArray.item = (i) => mimeTypeArray[i] || null;
+    mimeTypeArray.namedItem = (name) => {
+      const idx = fakeMimeTypes.findIndex((m) => m.type === name);
+      return idx >= 0 ? mimeTypeArray[idx] : null;
+    };
+    mimeTypeArray[Symbol.iterator] = function* () {
+      for (let i = 0; i < fakeMimeTypes.length; i++) yield mimeTypeArray[i];
+    };
+
+    Object.defineProperty(navigator, "mimeTypes", {
+      get: () => mimeTypeArray,
+      configurable: true,
+    });
+  } catch (_) {}
+
+  // 5. WebGL renderer/vendor spoofing
+  try {
+    const WEBGL_VENDOR = "Google Inc. (Google)";
+    const WEBGL_RENDERER =
+      "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)";
+
+    const originalGetParameter = WebGLRenderingContext.prototype.getParameter;
+    WebGLRenderingContext.prototype.getParameter = function (param) {
+      if (param === 0x9245) return WEBGL_VENDOR;
+      if (param === 0x9246) return WEBGL_RENDERER;
+      return originalGetParameter.call(this, param);
+    };
+
+    if (typeof WebGL2RenderingContext !== "undefined") {
+      const originalGetParameter2 = WebGL2RenderingContext.prototype.getParameter;
+      WebGL2RenderingContext.prototype.getParameter = function (param) {
+        if (param === 0x9245) return WEBGL_VENDOR;
+        if (param === 0x9246) return WEBGL_RENDERER;
+        return originalGetParameter2.call(this, param);
+      };
+    }
+  } catch (_) {}
+
+  // 6. Permissions API — notifications should return 'prompt', not 'denied'
+  try {
+    const originalQuery = Permissions.prototype.query;
+    Permissions.prototype.query = function (descriptor) {
+      if (descriptor && descriptor.name === "notifications") {
+        return Promise.resolve({ state: "prompt", onchange: null });
+      }
+      return originalQuery.call(this, descriptor);
+    };
+  } catch (_) {}
+
+  try {
+    Object.defineProperty(Notification, "permission", {
+      get: () => "default",
+      configurable: true,
+    });
+  } catch (_) {}
+
+  // 7. window.chrome.runtime — ensure it exists
+  try {
+    if (!window.chrome) {
+      window.chrome = {};
+    }
+    if (!window.chrome.runtime) {
+      window.chrome.runtime = {
+        connect: function () { return {}; },
+        sendMessage: function () {},
+      };
+    }
+  } catch (_) {}
+
+  // 8. Remove Chromedriver artifacts
+  try {
+    for (const prop of Object.getOwnPropertyNames(document)) {
+      if (prop.startsWith("$cdc_") || prop.startsWith("cdc_")) {
+        delete document[prop];
+      }
+    }
+  } catch (_) {}
+
+  // 9. navigator.hardwareConcurrency — ensure a reasonable value
+  try {
+    if (navigator.hardwareConcurrency < 4) {
+      Object.defineProperty(navigator, "hardwareConcurrency", {
+        get: () => 8,
+        configurable: true,
+      });
+    }
+  } catch (_) {}
+
+  // 10. navigator.deviceMemory — ensure a reasonable value
+  try {
+    if (!navigator.deviceMemory || navigator.deviceMemory < 4) {
+      Object.defineProperty(navigator, "deviceMemory", {
+        get: () => 8,
+        configurable: true,
+      });
+    }
+  } catch (_) {}
+
+  // 11. Patch iframe contentWindow.navigator to match parent
+  try {
+    const originalCreateElement = document.createElement.bind(document);
+    document.createElement = function (tagName, options) {
+      const el = originalCreateElement(tagName, options);
+      if (tagName.toLowerCase() === "iframe") {
+        el.addEventListener("load", function () {
+          try {
+            if (el.contentWindow && el.contentWindow.navigator) {
+              Object.defineProperty(el.contentWindow.navigator, "webdriver", {
+                get: () => false,
+                configurable: true,
+              });
+            }
+          } catch (_) {}
+        });
+      }
+      return el;
+    };
+  } catch (_) {}
+})();

clawbench/data/docker/Dockerfile

@@ -0,0 +1,51 @@
+FROM python:3.12-slim
+
+# Rootless Podman without subuid: only UID/GID 0 exist in the namespace.
+# 1) Disable apt sandbox (apt tries to switch to _apt user → setgroups fails)
+# 2) Wrap dpkg-statoverride so chown to non-root groups (e.g. messagebus) doesn't abort
+# 3) Restore real dpkg-statoverride after install
+RUN echo 'APT::Sandbox::User "root";' > /etc/apt/apt.conf.d/01disable-sandbox \
+    && mv /usr/bin/dpkg-statoverride /usr/bin/dpkg-statoverride.real \
+    && printf '#!/bin/sh\n/usr/bin/dpkg-statoverride.real "$@" 2>/dev/null || true\n' \
+    > /usr/bin/dpkg-statoverride && chmod +x /usr/bin/dpkg-statoverride \
+    && apt-get update && apt-get install -y --no-install-recommends \
+    chromium xvfb ffmpeg socat curl git x11vnc xclip \
+    libegl1 libgbm1 \
+    fonts-noto-color-emoji fonts-noto-cjk \
+    && mv /usr/bin/dpkg-statoverride.real /usr/bin/dpkg-statoverride \
+    && apt-get clean && rm -rf /var/lib/apt/lists/*
+
+# noVNC + websockify for human mode (browser-based VNC client)
+RUN git clone --depth 1 --branch v1.6.0 https://github.com/novnc/noVNC.git /opt/novnc \
+    && git clone --depth 1 --branch v0.13.0 https://github.com/novnc/websockify.git /opt/novnc/utils/websockify
+
+COPY --from=node:24-slim /usr/local/bin/node /usr/local/bin/node
+COPY --from=node:24-slim /usr/local/lib/node_modules /usr/local/lib/node_modules
+RUN ln -s /usr/local/lib/node_modules/npm/bin/npm-cli.js /usr/local/bin/npm \
+    && ln -s /usr/local/lib/node_modules/npm/bin/npx-cli.js /usr/local/bin/npx
+
+# Pin openclaw version — patched below for #47879 (--autoConnect → --browserUrl)
+RUN PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD=1 npm install -g openclaw@2026.3.13
+
+# Patch: replace --autoConnect with --browserUrl http://127.0.0.1:9222 in all dist files
+# See https://github.com/openclaw/openclaw/issues/47879
+RUN find /usr/local/lib/node_modules/openclaw/dist -name '*.js' -exec \
+    sed -i 's/"--autoConnect"/"--browserUrl","http:\/\/127.0.0.1:9222"/g' {} +
+
+COPY --from=ghcr.io/astral-sh/uv:0.11.6 /uv /usr/local/bin/uv
+
+WORKDIR /app
+COPY extension-server/ ./extension-server/
+RUN cd extension-server && UV_PYTHON_PREFERENCE=only-system uv sync
+
+COPY chrome-extension/ ./chrome-extension/
+
+COPY setup-openclaw.sh /setup-openclaw.sh
+RUN chmod +x /setup-openclaw.sh
+
+COPY entrypoint.sh /entrypoint.sh
+RUN chmod +x /entrypoint.sh
+
+EXPOSE 6080 7878 9223
+
+ENTRYPOINT ["/entrypoint.sh"]

clawbench/data/docker/entrypoint.sh

@@ -0,0 +1,394 @@
+#!/bin/bash
+set -e
+
+# Ensure /data exists for recording output and diagnostic logs
+mkdir -p /data
+
+# Start virtual display
+Xvfb :99 -screen 0 1920x1080x24 &
+export DISPLAY=:99
+sleep 1
+
+# Start the server
+cd /app/extension-server
+uv run uvicorn server:app --host 0.0.0.0 --port 7878 &
+sleep 1
+
+# Start Chrome with extension (realistic profile to reduce bot detection)
+mkdir -p /tmp/chrome-profile/Default
+
+cat > /tmp/chrome-profile/Default/Preferences <<'PREFS'
+{
+  "credentials_enable_service": false,
+  "profile": {
+    "password_manager_enabled": false,
+    "password_manager_leak_detection": false,
+    "name": "Default",
+    "created_by_version": "131.0.6778.139",
+    "content_settings": {
+      "exceptions": {}
+    }
+  },
+  "browser": {
+    "has_seen_welcome_page": true,
+    "check_default_browser": false,
+    "window_placement": {
+      "bottom": 1080,
+      "left": 0,
+      "maximized": false,
+      "right": 1920,
+      "top": 0,
+      "work_area_bottom": 1080,
+      "work_area_left": 0,
+      "work_area_right": 1920,
+      "work_area_top": 0
+    }
+  },
+  "dns_prefetching": {
+    "enabled": true
+  },
+  "safebrowsing": {
+    "enabled": true
+  },
+  "search": {
+    "suggest_enabled": true
+  },
+  "translate": {
+    "enabled": false
+  },
+  "intl": {
+    "accept_languages": "en-US,en"
+  },
+  "distribution": {
+    "import_bookmarks": false,
+    "skip_first_run_ui": true
+  }
+}
+PREFS
+
+cat > /tmp/chrome-profile/Default/Bookmarks <<'BOOKMARKS'
+{
+  "checksum": "b5f7e1a2c3d4e5f6a7b8c9d0e1f2a3b4",
+  "roots": {
+    "bookmark_bar": {
+      "children": [
+        {
+          "date_added": "13350000000000000",
+          "date_last_used": "0",
+          "guid": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
+          "name": "Google",
+          "type": "url",
+          "url": "https://www.google.com/"
+        },
+        {
+          "date_added": "13350000000000000",
+          "date_last_used": "0",
+          "guid": "b2c3d4e5-f6a7-8901-bcde-f12345678901",
+          "name": "YouTube",
+          "type": "url",
+          "url": "https://www.youtube.com/"
+        },
+        {
+          "date_added": "13350000000000000",
+          "date_last_used": "0",
+          "guid": "c3d4e5f6-a7b8-9012-cdef-123456789012",
+          "name": "Wikipedia",
+          "type": "url",
+          "url": "https://en.wikipedia.org/"
+        }
+      ],
+      "date_added": "13350000000000000",
+      "date_last_used": "0",
+      "date_modified": "13350000000000000",
+      "guid": "00000000-0000-4000-a000-000000000001",
+      "name": "Bookmarks bar",
+      "type": "folder"
+    },
+    "other": {
+      "children": [],
+      "date_added": "13350000000000000",
+      "date_last_used": "0",
+      "date_modified": "0",
+      "guid": "00000000-0000-4000-a000-000000000002",
+      "name": "Other bookmarks",
+      "type": "folder"
+    },
+    "synced": {
+      "children": [],
+      "date_added": "13350000000000000",
+      "date_last_used": "0",
+      "date_modified": "0",
+      "guid": "00000000-0000-4000-a000-000000000003",
+      "name": "Mobile bookmarks",
+      "type": "folder"
+    }
+  },
+  "version": 1
+}
+BOOKMARKS
+
+cat > /tmp/chrome-profile/'Local State' <<'LOCALSTATE'
+{
+  "browser": {
+    "enabled_labs_experiments": []
+  },
+  "profile": {
+    "info_cache": {
+      "Default": {
+        "active_time": 1710000000,
+        "is_consented_primary_account": false,
+        "name": "Person 1"
+      }
+    }
+  },
+  "user_experience_metrics": {
+    "reporting_enabled": false
+  }
+}
+LOCALSTATE
+
+chromium \
+  --window-size=1920,1080 \
+  --window-position=0,0 \
+  --no-first-run \
+  --disable-default-apps \
+  --no-sandbox \
+  --disable-infobars \
+  --disable-dev-shm-usage \
+  --disable-blink-features=AutomationControlled \
+  --use-gl=angle --use-angle=swiftshader \
+  --enable-unsafe-swiftshader \
+  --enable-webgl \
+  --password-store=basic \
+  --use-mock-keychain \
+  --disable-sync \
+  --disable-features=PasswordLeakDetection,PasswordManager \
+  --user-data-dir=/tmp/chrome-profile \
+  --remote-debugging-port=9222 \
+  --remote-debugging-address=127.0.0.1 \
+  --remote-allow-origins=* \
+  --load-extension=/app/chrome-extension \
+  --disable-extensions-except=/app/chrome-extension \
+  about:blank &
+
+# Forward CDP port to all interfaces
+sleep 2
+socat TCP-LISTEN:9223,fork,reuseaddr,bind=0.0.0.0 TCP:127.0.0.1:9222 &
+
+# Always start noVNC so users can watch the browser in any mode.
+# In human mode x11vnc also fires disconnect hooks for the watchdog.
+echo "Starting noVNC..."
+VNC_GONE_HOOK=""
+VNC_ACCEPT_HOOK=""
+if [ "$HUMAN_MODE" = "1" ]; then
+  VNC_GONE_HOOK="-gone touch /data/.vnc-disconnected"
+  VNC_ACCEPT_HOOK="-afteraccept rm -f /data/.vnc-disconnected"
+fi
+x11vnc -display :99 -nopw -shared -forever -rfbport 5900 -xkb \
+  $VNC_GONE_HOOK $VNC_ACCEPT_HOOK &
+sleep 1
+
+/opt/novnc/utils/novnc_proxy --vnc localhost:5900 --listen 6080 &
+sleep 1
+echo "============================================"
+echo "noVNC ready: http://localhost:6080/vnc.html"
+echo "============================================"
+
+# Human mode: wait for VNC disconnect or eval match
+if [ "$HUMAN_MODE" = "1" ]; then
+  echo "Human mode active."
+  if [ -n "$INSTRUCTION" ]; then
+    echo ""
+    echo "TASK: $INSTRUCTION"
+    echo ""
+  fi
+
+  # Human watchdog: no idle detection (humans pause to think)
+  MAX_WAIT=${TIME_LIMIT_S:-1800}
+  ELAPSED=0
+  DISCONNECT_WAIT=0
+  STOP_REASON=""
+
+  while [ "$ELAPSED" -lt "$MAX_WAIT" ]; do
+    sleep 5
+    ELAPSED=$((ELAPSED + 5))
+
+    # Check if eval interceptor matched
+    if [ -f /data/.stop-requested ]; then
+      echo "Stop requested by server (eval matched)."
+      STOP_REASON="eval_matched"
+      break
+    fi
+
+    # Check if VNC client disconnected (with 15s grace period for reconnect)
+    if [ -f /data/.vnc-disconnected ]; then
+      DISCONNECT_WAIT=$((DISCONNECT_WAIT + 5))
+      if [ "$DISCONNECT_WAIT" -ge 15 ]; then
+        echo "VNC client disconnected for ${DISCONNECT_WAIT}s, assuming done."
+        STOP_REASON="vnc_disconnected"
+        break
+      fi
+    else
+      DISCONNECT_WAIT=0
+    fi
+  done
+
+  if [ -z "$STOP_REASON" ]; then
+    echo "Time limit (${MAX_WAIT}s) exceeded."
+    STOP_REASON="time_limit_exceeded"
+  fi
+
+  echo "$STOP_REASON" > /data/.stop-reason
+
+  # Finalize bookkeeping (eval promotion, etc.) — recording keeps running
+  curl -sf -X POST http://localhost:7878/api/stop || true
+  rm -f /data/.stop-requested /data/.vnc-disconnected
+
+  # Grace period: keep recording for 15s to capture end state
+  echo "Human finished, recording grace period (15s)..."
+  sleep 15
+
+  # Stop recording
+  echo "Stopping recording..."
+  curl -sf -X POST http://localhost:7878/api/stop-recording || true
+  sleep 2
+  echo "Done."
+  exit 0
+fi
+
+# If no INSTRUCTION provided, skip OpenClaw and keep container alive for external use
+if [ -z "$INSTRUCTION" ]; then
+  echo "No INSTRUCTION set, running in manual mode (no OpenClaw agent)."
+  wait -n
+  exit 0
+fi
+
+# Generate a shared gateway token so gateway and agent can authenticate
+OPENCLAW_GATEWAY_TOKEN="$(head -c 32 /dev/urandom | od -A n -t x1 | tr -d ' \n')"
+export OPENCLAW_GATEWAY_TOKEN
+
+# Generate OpenClaw config from env vars
+/setup-openclaw.sh
+
+# Copy /my-info/ into the OpenClaw workspace so the agent can access it via ./my-info/
+WORKSPACE=/root/workspace
+mkdir -p "$WORKSPACE"
+if [ -d /my-info ]; then
+  cp -r /my-info "$WORKSPACE/my-info"
+  echo "Copied /my-info to $WORKSPACE/my-info"
+fi
+
+# Wait for Chrome CDP to be ready
+echo "Waiting for Chrome CDP..."
+for i in $(seq 1 30); do
+  if curl -sf http://127.0.0.1:9222/json/version > /dev/null 2>&1; then
+    echo "Chrome CDP ready"
+    break
+  fi
+  if [ "$i" -eq 30 ]; then
+    echo "Chrome CDP not ready after 30s, aborting"
+    echo "chrome_cdp_timeout" > /data/.stop-reason
+    exit 1
+  fi
+  sleep 1
+done
+
+# Start OpenClaw gateway (log to /data for post-mortem debugging)
+openclaw gateway run > /data/gateway.log 2>&1 &
+GATEWAY_PID=$!
+sleep 3
+
+# Check gateway is alive
+if ! kill -0 $GATEWAY_PID 2>/dev/null; then
+  echo "ERROR: OpenClaw gateway died on startup. Log:"
+  cat /data/gateway.log
+  echo "gateway_failed" > /data/.stop-reason
+  exit 1
+fi
+echo "OpenClaw gateway running (pid=$GATEWAY_PID)"
+
+# Run the agent from workspace directory (where my-info/ lives)
+echo "Starting OpenClaw agent from $WORKSPACE..."
+TIMEOUT_MS=$(( ${TIME_LIMIT_S:-1800} * 1000 ))
+AGENT_CMD=(openclaw agent --session-id clawbench --message "$INSTRUCTION" --thinking "${THINKING_LEVEL:-medium}" --timeout "$TIMEOUT_MS" --local)
+echo "Agent command: ${AGENT_CMD[*]}"
+if [ -n "$TEMPERATURE" ]; then
+  AGENT_CMD+=(--temperature "$TEMPERATURE")
+fi
+if [ -n "$MAX_TOKENS" ]; then
+  AGENT_CMD+=(--max-tokens "$MAX_TOKENS")
+fi
+cd "$WORKSPACE"
+"${AGENT_CMD[@]}" > /data/agent.log 2>&1 &
+AGENT_PID=$!
+
+# Watchdog: wait for agent activity, then detect idle (no new actions for 300s)
+IDLE_THRESHOLD=300
+MAX_WAIT=${TIME_LIMIT_S:-1800}
+ELAPSED=0
+LAST_SIZE=0
+IDLE=0
+STOP_REASON=""
+
+while kill -0 $AGENT_PID 2>/dev/null && [ "$ELAPSED" -lt "$MAX_WAIT" ]; do
+  sleep 5
+  ELAPSED=$((ELAPSED + 5))
+
+  # Check if server requested stop (eval interceptor matched)
+  if [ -f /data/.stop-requested ]; then
+    echo "Stop requested by server (eval matched), killing agent."
+    STOP_REASON="eval_matched"
+    break
+  fi
+
+  CURRENT_SIZE=$(wc -c < /data/actions.jsonl 2>/dev/null || echo 0)
+
+  if [ "$CURRENT_SIZE" -gt 0 ] && [ "$CURRENT_SIZE" -eq "$LAST_SIZE" ]; then
+    IDLE=$((IDLE + 5))
+    if [ "$IDLE" -ge "$IDLE_THRESHOLD" ]; then
+      echo "Agent idle for ${IDLE_THRESHOLD}s, assuming done."
+      STOP_REASON="agent_idle"
+      break
+    fi
+  else
+    IDLE=0
+  fi
+  LAST_SIZE=$CURRENT_SIZE
+done
+
+# Determine stop reason if not set (loop exited without breaking)
+if [ -z "$STOP_REASON" ]; then
+  if ! kill -0 $AGENT_PID 2>/dev/null; then
+    STOP_REASON="agent_exited"
+  else
+    echo "Time limit (${MAX_WAIT}s) exceeded, killing agent."
+    STOP_REASON="time_limit_exceeded"
+  fi
+fi
+
+echo "$STOP_REASON" > /data/.stop-reason
+
+# Kill all openclaw processes
+kill $AGENT_PID 2>/dev/null || true
+kill $GATEWAY_PID 2>/dev/null || true
+pkill -f "openclaw" 2>/dev/null || true
+sleep 2
+
+# Copy OpenClaw session transcript to /data
+cp /root/.openclaw/agents/main/sessions/clawbench.jsonl /data/agent-messages.jsonl 2>/dev/null || true
+
+# Finalize bookkeeping (eval promotion, etc.) — recording keeps running
+curl -sf -X POST http://localhost:7878/api/stop || true
+
+# Clean up internal marker (created by /api/stop)
+rm -f /data/.stop-requested
+
+# Grace period: keep recording for 15s after agent is killed to capture end result
+echo "Agent finished, recording grace period (15s)..."
+sleep 15
+
+# Stop recording
+echo "Stopping recording..."
+curl -sf -X POST http://localhost:7878/api/stop-recording || true
+sleep 2
+echo "Done."