diff --git a/RUN.md b/RUN.md
index 9c77b71..e8c4c7a 100644
--- a/RUN.md
+++ b/RUN.md
@@ -387,6 +387,110 @@ sudo -u zcbot docker rm -f zcbot-sandbox-$USER_ID
 Step 4 引入 egress proxy 后,完整 5 条红队用例(metadata / loopback / 跨 user / nohup
 残留 / allowlist 外 403)进 `tests/test_sandbox_redteam.py` 自动化跑。
 
+### Stage C 完整部署速查
+
+首次部署 / 改了 init.sh / Dockerfile / requirements.txt → 重 build:
+
+```bash
+cd /home/lighthouse/zcbot
+docker build -t zcbot-sandbox:latest -f deploy/sandbox/Dockerfile \
+    --build-arg HOST_UID=$(id -u) --build-arg HOST_GID=$(id -g) \
+    --build-arg APT_MIRROR=http://mirrors.cloud.tencent.com \
+    --build-arg PIP_INDEX_URL=https://mirrors.cloud.tencent.com/pypi/simple/ \
+    --build-arg NPM_REGISTRY=https://mirrors.cloud.tencent.com/npm/ \
+    .
+```
+
+只改 host 侧 docker run flag / pool.py(没动 Dockerfile / init.sh)→ 不用重 build,
+清旧容器 + 重启 web 即可:
+
+```bash
+docker rm -f $(docker ps -aq -f label=zcbot.product=sandbox) 2>/dev/null
+systemctl restart zcbot
+```
+
+跑 migration(新增 / 修改了 db/migrations/versions/* / models.py):
+
+```bash
+.venv/bin/python main.py db upgrade head
+```
+
+启用 docker backend 重启 web(确保 `.env` 有 `ZCBOT_SANDBOX_BACKEND=docker` 或 systemd
+unit 已设):
+
+```bash
+systemctl restart zcbot
+journalctl -u zcbot -e --no-pager | tail -30   # 看 startup log
+# 应见:
+#   [startup] [warn] fs quota: ext4 on ... (dogfood 阶段 warn 正常)
+#   [startup] swept N stale sandbox container(s)
+#   [disk_scanner] initial scan: N user(s)
+#   INFO: Uvicorn running on 0.0.0.0:8765
+```
+
+### 日常运维 / 验证
+
+```bash
+# 当前活跃 sandbox 容器
+docker ps --filter label=zcbot.product=sandbox \
+    --format 'table {{.Names}}\t{{.Status}}\t{{.RunningFor}}'
+
+# 某容器跑了什么(看 docker run 完整参数 + env + mount)
+SBC=$(docker ps -q -f label=zcbot.product=sandbox | head -1)
+docker inspect $SBC --format '{{json .Config.Env}}' | python3 -m json.tool
+docker inspect $SBC --format '{{json .HostConfig.Binds}}'
+
+# 容器内 DNS + 网络验证(应见 nameserver 8.8.8.8 / 114.114.114.114)
+docker exec $SBC cat /etc/resolv.conf
+docker exec $SBC getent hosts www.baidu.com
+docker exec $SBC curl -sI -m 5 https://www.baidu.com | head -1
+
+# iptables 规则(应见红线段 DROP + 127.0.0.11:53 ACCEPT 例外)
+docker exec $SBC iptables -L OUTPUT -n --line-numbers
+
+# 容器内 zcbot user 与 host 对齐(uid 应一致)
+docker exec $SBC id
+id   # host 上跑
+
+# 容器资源占用
+docker stats --no-stream $SBC
+
+# 看本会话 sandbox 启动 stderr(init.sh 的 [init] ... 行)
+docker logs $SBC
+
+# 杀某 user 容器(强制下次 ensure 重建,可用于 hotfix 后切镜像)
+docker rm -f zcbot-sandbox-<user-uuid>
+
+# 清所有 sandbox 容器(reaper 兜底,不影响 web 进程)
+docker rm -f $(docker ps -aq -f label=zcbot.product=sandbox) 2>/dev/null
+```
+
+磁盘配额查看(0008 表):
+
+```bash
+# 当前所有 user 用量
+psql "$ZCBOT_DB_URL" -c "
+    SELECT user_id, bytes_used/1024/1024 AS mb, file_count, scanned_at
+    FROM user_disk_usage ORDER BY bytes_used DESC;"
+
+# 强制扫描某 user(改 yaml interval 太久不耐烦时)
+.venv/bin/python -c "
+from pathlib import Path
+from uuid import UUID
+from core.storage.disk_quota import scan_user_dir, upsert_user_usage
+uid = UUID('<your-user-uuid>')
+b, c = scan_user_dir(Path('/home/lighthouse/zcbot/workspace/users') / str(uid))
+upsert_user_usage(uid, b, c)
+print(f'{b/1024/1024:.1f} MB / {c} files')"
+```
+
+DNS / resolv.conf 文件位置(host 侧):
+
+```bash
+# 由 SandboxPool 在首次 _docker_run 时生成,每次重建容器会覆盖刷新
+cat /home/lighthouse/zcbot/workspace/.sandbox/resolv.conf
+```
+
 ### 部署前置对账
 
 切 `ZCBOT_SANDBOX_BACKEND=docker` 之前跑一次: