feat: 增加 Sub2API 用户反馈只读追踪

docs/reference/cli.md

@@ -164,7 +164,7 @@ PipelineRun 失败或长时间未完成时，先按定点 `control-plane status
 - `hwlab g14 git-mirror status|apply|sync|flush [--dry-run|--confirm]` 是 `devops-infra` git mirror/relay 的受控维护入口：`apply` 渲染并 server-side apply `devops-infra/git-mirror.yaml`，同时删除遗留 `git-mirror-hwlab-sync` CronJob；`sync` 创建一次性 manual Job，把 GitHub allowlist refs 拉入本地 mirror；`flush` 创建一次性 manual Job，把本地 `v0.2-gitops` 快进推回 GitHub。
   `status` 返回 read/write URL、last sync/write/flush、本地 ref、GitHub staging ref 和 pending flush 状态，并在 `cache.summary` 给出 `localV02`、`localGitops`、`githubGitops`、`pendingFlush`、`flushNeeded`、`githubInSync` 和下一条受控 `flushCommand`。confirmed `sync` 和 `flush` 默认创建 `.state/jobs/` 异步 job 并立刻返回可查询状态，只有现场同步调试才显式加 `--wait`；mirror 不设置 CronJob。
   如果 PipelineRun 的 `gitops-promote` 阶段报 git mirror 控制面漂移或 refs 不一致，先执行 `hwlab g14 git-mirror apply --confirm` 重新应用当前 `devops-infra/git-mirror.yaml` hook/ConfigMap，再执行 `hwlab g14 git-mirror sync --confirm --wait` 复核 refs；失败的同名 PipelineRun 只能通过 `hwlab g14 control-plane cleanup-runs --lane <lane> --pipeline-run <name> --confirm` 受控清理后重试，不要用原生 `kubectl delete` 或手工改 mirror hook。修复后仍必须用 `control-plane status --pipeline-run <name>` 和 `git-mirror status` 分别确认 runtime closeout 与 GitHub flush。
-- `platform-infra sub2api|langbot|n8n|wechat-archive ...` 是 `platform-infra` namespace 内平台公共服务和公共工作流的受控入口；`sub2api` 支持 `plan|apply|status|validate|codex-pool`，`langbot` 和 `n8n` 支持各自 YAML-controlled `plan|apply|status|logs|validate` 等公共服务操作，`wechat-archive` 支持 `plan|apply|status|validate|pull`，用 `config/platform-infra/wechat-archive.yaml` 声明 LangBot/n8n/Baidu Netdisk 归档链路，以及 D601 Windows 隔离 PC 微信 + WeChatFerry host + D601 k3s 只读 collector 的 personal WeChat upstream；collector 必须复用 D601 已有 `platform-infra` namespace，`createNamespace=false`。`--target` 选择运行目标，默认 `G14` 为 active runtime，`D601` 为同一 YAML 控制的 standby predeploy target。镜像版本和 target 边界由 `config/platform-infra/*.yaml` 控制，Codex 上游池、统一 API key Secret、FRP 公网端口和 master `~/.codex` 消费端由 `config/platform-infra/sub2api-codex-pool.yaml` 控制；完整 Sub2API 日常部署、上游增删、FRP 暴露、local Codex 配置、验收和排障步骤统一见 `$unidesk-sub2api`（`.agents/skills/unidesk-sub2api/SKILL.md`）。`docs/reference/platform-infra.md` 保留 namespace、YAML-first、路由、Secret 脱敏、PK01 Caddy+FRP 和探针开发边界。
+- `platform-infra sub2api|langbot|n8n|wechat-archive ...` 是 `platform-infra` namespace 内平台公共服务和公共工作流的受控入口；`sub2api` 支持 `plan|apply|status|validate|codex-pool`，其中用户反馈一键只读追踪见 `docs/reference/sub2api-user-feedback.md`；`langbot` 和 `n8n` 支持各自 YAML-controlled `plan|apply|status|logs|validate` 等公共服务操作，`wechat-archive` 支持 `plan|apply|status|validate|pull`，用 `config/platform-infra/wechat-archive.yaml` 声明 LangBot/n8n/Baidu Netdisk 归档链路，以及 D601 Windows 隔离 PC 微信 + WeChatFerry host + D601 k3s 只读 collector 的 personal WeChat upstream；collector 必须复用 D601 已有 `platform-infra` namespace，`createNamespace=false`。`--target` 选择运行目标，默认 `G14` 为 active runtime，`D601` 为同一 YAML 控制的 standby predeploy target。镜像版本和 target 边界由 `config/platform-infra/*.yaml` 控制，Codex 上游池、统一 API key Secret、FRP 公网端口和 master `~/.codex` 消费端由 `config/platform-infra/sub2api-codex-pool.yaml` 控制；完整 Sub2API 日常部署、上游增删、FRP 暴露、local Codex 配置、验收和排障步骤统一见 `$unidesk-sub2api`（`.agents/skills/unidesk-sub2api/SKILL.md`）。`docs/reference/platform-infra.md` 保留 namespace、YAML-first、路由、Secret 脱敏、PK01 Caddy+FRP 和探针开发边界。
 - `platform-infra observability plan|apply|status|validate|trace --target <node>` 是 `platform-infra` 内 OTel Collector 和 Tempo trace backend 的受控入口，`--target` 必须选择 trace 所属运行面节点；具体 trace 查询、噪声压制、Code Agent/AgentRun 排障和“先改进 OTel”的操作面统一见 `$unidesk-otel`（`.agents/skills/unidesk-otel/SKILL.md`）。
 - `secrets plan|sync|status --config config/secrets-distribution.yaml --scope platform-infra` 是平台服务本地 Secret sourceRef 到 Kubernetes Secret key 的受控下发入口。`plan` 只读展示 sourceRef、必需 key、目标 Secret/key、missingKeys 和 fingerprint；`sync --confirm` 只按 YAML 声明创建允许生成的本地 key 并下发到声明的目标 Secret；`status` 只验证 live Secret key presence。该入口禁止从 live pod 或 Kubernetes Secret 反推密码、API key、JWT secret、n8n encryption key 或 `DATABASE_URL`，输出也不得打印 base64、解码值或远端 raw transcript；即使显式 `--raw` 也只返回脱敏 summary 和 raw omission 标记。LangBot/n8n Secret 轮换和缺 key 修复规则见 `docs/reference/platform-infra.md#secret-distribution-boundary`。
 - `hwlab g14 observability status|apply|query|targets|boundary|closeout [--lane v02] [--promql <expr>] [--expect-count N] [--expect-value V] [--dry-run|--confirm]` 是 G14 `devops-infra` 共享监控基础设施和 HWLAB v0.2 监控 closeout 的受控入口。`apply` 固定安装 Prometheus Operator `v0.91.0`、Prometheus `v3.12.0`、Prometheus 发现 RBAC、`devops-infra` 内 Prometheus 实例和 ClusterIP query Service，并给被允许发现的 workload namespace 打低风险 label；它不把 Prometheus、Grafana 或 Alertmanager 部署到 `hwlab-v02`，也不接管 HWLAB runtime Deployment/Service。`status` 只读汇总 CRD、operator Deployment、Prometheus CR/pod/service、`hwlab-v02` ServiceMonitor/PrometheusRule 和 bounded `up` 查询；`query` 只通过 Kubernetes service proxy 查询 Prometheus，支持 `--expect-count` / `--expect-value` 输出 `assertion`、bad values 和 missing/extra series；`targets` 汇总 ServiceMonitor/PrometheusRule、metrics sidecar readiness/restart、三层指标值和 `metrics.k8s.io` 当前 CPU/内存资源快照；`boundary` 验证 workload namespace 没有 Prometheus/Alertmanager，并对 `19666/19667` 公网 `/metrics` 做负向验证；`closeout` 聚合平台 ready、scrape reachable、sidecar serving、business health probe、resource snapshot、namespace boundary 和 public metrics exposure 语义结论。长期边界见 `docs/reference/g14-observability-infra.md`。

docs/reference/sub2api-user-feedback.md

@@ -0,0 +1,61 @@
+# Sub2API 用户反馈只读追踪
+
+## 入口
+
+- 按用户邮箱或正整数用户 ID 查询：
+
+  ```bash
+  bun scripts/cli.ts platform-infra sub2api codex-pool feedback \
+    --target PK01 \
+    --user <email-or-id>
+  ```
+
+- 默认窗口为 `30m`；可显式选择 `5m|30m|1h|6h|24h|7d|30d`。
+- 默认输出 Kubernetes 列表风格紧凑文本；只有显式 `--json` 才输出完整脱敏 JSON。
+- 请求表固定返回 10 条，并输出 `returned`、`total`、`moreAvailable`、`nextPageToken`；下一页使用 CLI 给出的完整 `Next` 命令，不提供手工 `limit`。
+- `REQUEST ID INDEX` 输出当前页完整、可复制的 request ID；主表使用页内序号避免重复宽字段。
+- 按默认表格中的 client request ID 或内部 request ID 下钻：
+
+  ```bash
+  bun scripts/cli.ts platform-infra sub2api codex-pool feedback \
+    --target PK01 \
+    --user <email-or-id> \
+    --request-id <client-or-internal-id>
+  ```
+
+- 稳定游标下一页：
+
+  ```bash
+  bun scripts/cli.ts platform-infra sub2api codex-pool feedback \
+    --target PK01 \
+    --user <email-or-id> \
+    --window 24h \
+    --page-token <nextPageToken>
+  ```
+
+## 证据组合
+
+- 用户与 API Key：使用 Sub2API 原生 `admin users` 和用户 API Key 列表，输出邮箱、名称、分组与当前并发，不输出 Key 值。
+- 请求时间线：自动分页读取窗口内全部 `ops requests`，逐项显示相邻请求空窗、模型、状态、耗时、账号和 proxy。
+- ID 关联：一次读取 `ops system-logs`，在本地关联 client request ID 与内部 request ID；指定 `--request-id` 时披露关联日志。
+- 实时状态：组合用户、账号和分组并发，区分无入站证据、排队、在途与已完成。
+- 基础设施：组合账号可用性、账号当前状态、proxy 状态和 Sub2API 组件健康摘要。
+
+## 归因边界
+
+- `client-before-submit-or-no-inbound-evidence` 表示窗口内没有 Sub2API 入站记录且实时并发为零，不等价于证明客户端没有尝试调用。
+- `sub2api-queue` 需要实时等待队列证据；`in-flight` 需要实时并发证据。
+- `upstream-first-byte-stream`、`proxy-egress` 和 `client-connection` 只根据原生 Ops 日志、账号、proxy 与基础设施证据分类，不执行探针或写操作。
+- 无足够分层证据时返回 `sub2api-or-upstream-unknown`，禁止扩大结论。
+
+## 脱敏与只读
+
+- 入口只调用 GET 类 admin/ops 接口、管理员登录和运行组件健康读取。
+- 不调用账号测试、探针、调度、取消、清理、更新或其他运行面写接口。
+- JSON 和文本均不披露 API Key、管理员 token、账号 credentials、proxy 用户名、密码或其他 Secret。
+- 所有分页由 CLI 自动完成，不提供手工 `limit` 作为输出规避手段。
+
+## 调用量
+
+- issue `#2000` 的人工调查包含用户、请求、系统日志、两种 request ID trace、账号、proxy、进程/端口和 journal 共 9 次工具入口。
+- 本入口将用户侧工具调用收敛为 1 次，减少 8 次，即 `88.9%`；内部只读 API 来源在 JSON `sources` 字段披露。

scripts/src/platform-infra-sub2api-codex/feedback.ts

@@ -0,0 +1,593 @@
+import type { UniDeskConfig } from "../config";
+import type { RenderedCliResult } from "../output";
+
+import { readCodexPoolConfig } from "./config";
+import { capture, compactCapture, parseJsonOutput } from "./remote";
+import { renderedCliResult, renderTable, shorten, shortIso, textValue } from "./render";
+import { codexPoolRuntimeTarget } from "./runtime-target";
+import type { CodexPoolConfig, CodexPoolRuntimeTarget, FeedbackOptions } from "./types";
+
+function pyJson(value: unknown): string {
+  return `json.loads(${JSON.stringify(JSON.stringify(value))})`;
+}
+
+export async function codexPoolFeedback(config: UniDeskConfig, options: FeedbackOptions): Promise<Record<string, unknown> | RenderedCliResult> {
+  const pool = readCodexPoolConfig();
+  const target = codexPoolRuntimeTarget(options.targetId);
+  const result = await capture(config, target.route, ["sh"], feedbackScript(pool, options, target));
+  const report = parseJsonOutput(result.stdout);
+  const ok = result.exitCode === 0 && report?.ok === true;
+  const remote = compactCapture(result, { full: result.exitCode !== 0 || report === null });
+  if (options.json) {
+    return {
+      ok,
+      action: "platform-infra-sub2api-codex-pool-feedback",
+      target: { id: target.id, route: target.route, runtimeMode: target.runtimeMode },
+      report,
+      remote,
+      valuesPrinted: false,
+    };
+  }
+  return renderedCliResult(ok, "platform-infra sub2api codex-pool feedback", renderFeedbackReport(report, options, remote));
+}
+
+function feedbackScript(pool: CodexPoolConfig, options: FeedbackOptions, target: CodexPoolRuntimeTarget): string {
+  return `
+set -u
+python3 - <<'PY'
+import base64
+import json
+import re
+import subprocess
+from datetime import datetime, timezone, timedelta
+from urllib.parse import quote, urlencode
+
+RUNTIME_MODE = ${pyJson(target.runtimeMode)}
+NAMESPACE = ${pyJson(target.namespace)}
+APP_SECRET_NAME = ${pyJson(target.appSecretName)}
+ADMIN_EMAIL_DEFAULT = ${pyJson(pool.adminEmailDefault)}
+HOST_DOCKER_APP_PORT = ${pyJson(target.hostDockerAppPort)}
+HOST_DOCKER_ENV_PATH = ${pyJson(target.hostDockerEnvPath)}
+USER_SELECTOR = ${pyJson(options.user)}
+WINDOW = ${pyJson(options.window)}
+REQUEST_SELECTOR = ${pyJson(options.requestId)}
+PAGE_TOKEN = ${pyJson(options.pageToken)}
+APP_CONTAINER = "sub2api-app"
+PAGE_SIZE = 10
+
+def run(command, input_bytes=None):
+    return subprocess.run(command, input=input_bytes, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+
+def text(value, limit=1000):
+    if isinstance(value, bytes):
+        value = value.decode("utf-8", errors="replace")
+    return str(value or "")[-limit:]
+
+def docker(args):
+    proc = run(["docker", *args])
+    if proc.returncode == 0:
+        return proc
+    fallback = run(["sudo", "-n", "docker", *args])
+    return fallback if fallback.returncode == 0 else proc
+
+def read_host_env():
+    if RUNTIME_MODE != "host-docker":
+        return {}
+    try:
+        with open(HOST_DOCKER_ENV_PATH, "r", encoding="utf-8") as handle:
+            lines = handle.read().splitlines()
+    except PermissionError:
+        proc = run(["sudo", "-n", "cat", HOST_DOCKER_ENV_PATH])
+        if proc.returncode != 0:
+            raise RuntimeError("read host-docker env failed: " + text(proc.stderr))
+        lines = proc.stdout.decode("utf-8", errors="replace").splitlines()
+    values = {}
+    for line in lines:
+        stripped = line.strip()
+        if not stripped or stripped.startswith("#") or "=" not in stripped:
+            continue
+        key, value = stripped.split("=", 1)
+        value = value.strip()
+        if len(value) >= 2 and value[0] == value[-1] and value[0] in ("'", '"'):
+            value = value[1:-1]
+        values[key.strip()] = value
+    return values
+
+def runtime_secret(key):
+    if RUNTIME_MODE == "host-docker":
+        return read_host_env().get(key)
+    proc = run(["kubectl", "-n", NAMESPACE, "get", "secret", APP_SECRET_NAME, "-o", "json"])
+    if proc.returncode != 0:
+        raise RuntimeError("read app secret failed: " + text(proc.stderr))
+    raw = (json.loads(proc.stdout.decode("utf-8")).get("data") or {}).get(key)
+    return base64.b64decode(raw).decode("utf-8") if raw else None
+
+def runtime_config(key):
+    if RUNTIME_MODE == "host-docker":
+        return read_host_env().get(key)
+    proc = run(["kubectl", "-n", NAMESPACE, "get", "configmap", "sub2api-config", "-o", "json"])
+    if proc.returncode != 0:
+        return None
+    return (json.loads(proc.stdout.decode("utf-8")).get("data") or {}).get(key)
+
+def parse_curl(proc):
+    stdout = proc.stdout.decode("utf-8", errors="replace")
+    marker = "\\n__HTTP_CODE__:"
+    position = stdout.rfind(marker)
+    if position < 0:
+        return {"ok": False, "httpStatus": 0, "json": None, "message": text(proc.stderr)}
+    body = stdout[:position]
+    try:
+        status = int(stdout[position + len(marker):].strip()[-3:])
+    except Exception:
+        status = 0
+    try:
+        parsed = json.loads(body) if body.strip() else None
+    except Exception:
+        parsed = None
+    message = parsed.get("message") if isinstance(parsed, dict) else text(body)
+    return {"ok": proc.returncode == 0 and 200 <= status < 300, "httpStatus": status, "json": parsed, "message": message}
+
+def curl_api(method, path, bearer=None, payload=None):
+    body = b"" if payload is None else json.dumps(payload, separators=(",", ":")).encode("utf-8")
+    script = r'''set -eu
+method="$1"
+url="$2"
+token="\${3:-}"
+tmp="$(mktemp)"
+trap 'rm -f "$tmp"' EXIT
+cat > "$tmp"
+args=""
+if [ -n "$token" ]; then args="Authorization: Bearer $token"; fi
+if [ "$method" = GET ] && [ ! -s "$tmp" ]; then
+  if [ -n "$args" ]; then curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -H "$args" "$url"; else curl -sS -w '\\n__HTTP_CODE__:%{http_code}' "$url"; fi
+else
+  if [ -n "$args" ]; then curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" -H 'Content-Type: application/json' -H "$args" --data-binary @"$tmp" "$url"; else curl -sS -w '\\n__HTTP_CODE__:%{http_code}' -X "$method" -H 'Content-Type: application/json' --data-binary @"$tmp" "$url"; fi
+fi'''
+    if RUNTIME_MODE == "host-docker":
+        proc = run(["sh", "-c", script, "sh", method, f"http://127.0.0.1:{HOST_DOCKER_APP_PORT}{path}", bearer or ""], body)
+    else:
+        proc = run(["kubectl", "-n", NAMESPACE, "exec", "deploy/sub2api", "--", "sh", "-c", script, "sh", method, f"http://127.0.0.1:8080{path}", bearer or ""], body)
+    return parse_curl(proc)
+
+def api_data(method, path, token=None, payload=None, label=None):
+    response = curl_api(method, path, token, payload)
+    parsed = response.get("json")
+    code = parsed.get("code") if isinstance(parsed, dict) else None
+    if not response.get("ok") or (code is not None and code != 0):
+        raise RuntimeError(f"{label or path} failed: http={response.get('httpStatus')} message={response.get('message')}")
+    return parsed.get("data") if isinstance(parsed, dict) and "data" in parsed else parsed
+
+def find_token(value):
+    if isinstance(value, dict):
+        for key in ("access_token", "token"):
+            if isinstance(value.get(key), str) and value.get(key):
+                return value[key]
+        for item in value.values():
+            token = find_token(item)
+            if token:
+                return token
+    return None
+
+def login():
+    email = runtime_config("ADMIN_EMAIL") or ADMIN_EMAIL_DEFAULT
+    password = runtime_secret("ADMIN_PASSWORD")
+    if not password:
+        raise RuntimeError("ADMIN_PASSWORD missing")
+    token = find_token(api_data("POST", "/api/v1/auth/login", payload={"email": email, "password": password}, label="admin login"))
+    if not token:
+        raise RuntimeError("admin login returned no token")
+    return email, token
+
+def items(data):
+    if isinstance(data, list):
+        return data
+    if isinstance(data, dict):
+        if isinstance(data.get("items"), list):
+            return data["items"]
+        for key in ("users", "api_keys", "keys", "groups", "accounts", "proxies", "logs"):
+            if isinstance(data.get(key), list):
+                return data[key]
+    return []
+
+def page_all(token, path, query=None, page_size=100):
+    query = dict(query or {})
+    collected = []
+    page = 1
+    while True:
+        params = {**query, "page": page, "page_size": page_size}
+        separator = "&" if "?" in path else "?"
+        data = api_data("GET", path + separator + urlencode(params), token, label=path)
+        batch = items(data)
+        collected.extend(batch)
+        total = data.get("total") if isinstance(data, dict) else None
+        if not batch or (isinstance(total, int) and len(collected) >= total) or len(batch) < page_size:
+            break
+        page += 1
+    return collected
+
+def exact_user(token):
+    if USER_SELECTOR.isdigit() and int(USER_SELECTOR) > 0:
+        user = api_data("GET", f"/api/v1/admin/users/{int(USER_SELECTOR)}", token, label="get user")
+        return user if isinstance(user, dict) else None
+    candidates = page_all(token, "/api/v1/admin/users", {"search": USER_SELECTOR}, 100)
+    exact = [user for user in candidates if isinstance(user, dict) and str(user.get("email") or "").lower() == USER_SELECTOR.lower()]
+    if len(exact) > 1:
+        raise RuntimeError("email matched multiple users")
+    return exact[0] if exact else None
+
+def safe_call(errors, name, callback, fallback):
+    try:
+        return callback()
+    except Exception as exc:
+        errors.append({"source": name, "error": redact_text(str(exc))})
+        return fallback
+
+def redact_text(value):
+    value = str(value or "")
+    value = re.sub(r"(?i)(bearer\\s+)[A-Za-z0-9._~+\\-/=]+", r"\\1<redacted>", value)
+    value = re.sub(r"(?i)((?:api[_-]?key|token|password)\\s*[=:]\\s*)[^\\s,;]+", r"\\1<redacted>", value)
+    value = re.sub(r"\\bsk-[A-Za-z0-9_-]{8,}\\b", "sk-<redacted>", value)
+    return value[:500]
+
+def iso_epoch(value):
+    if not isinstance(value, str) or not value:
+        return None
+    try:
+        match = re.match(r"^(\\d{4}-\\d\\d-\\d\\d[T ]\\d\\d:\\d\\d:\\d\\d)(?:\\.(\\d+))?(Z|[+-]\\d\\d:?\\d\\d)$", value)
+        if not match:
+            return None
+        base = datetime.strptime(match.group(1).replace(" ", "T"), "%Y-%m-%dT%H:%M:%S")
+        fraction = int((match.group(2) or "0")[:6].ljust(6, "0"))
+        zone = match.group(3)
+        if zone == "Z":
+            offset = timezone.utc
+        else:
+            sign = 1 if zone[0] == "+" else -1
+            digits = zone[1:].replace(":", "")
+            offset = timezone(sign * timedelta(hours=int(digits[:2]), minutes=int(digits[2:])))
+        return base.replace(microsecond=fraction, tzinfo=offset).timestamp()
+    except Exception:
+        return None
+
+def encode_page_token(row):
+    payload = json.dumps({"v": 1, "at": row.get("at"), "requestId": row.get("requestId")}, separators=(",", ":")).encode("utf-8")
+    return base64.urlsafe_b64encode(payload).decode("ascii").rstrip("=")
+
+def decode_page_token(value):
+    if not isinstance(value, str) or not value:
+        return None
+    try:
+        padded = value + "=" * (-len(value) % 4)
+        parsed = json.loads(base64.urlsafe_b64decode(padded.encode("ascii")).decode("utf-8"))
+    except Exception:
+        raise RuntimeError("invalid page token")
+    if not isinstance(parsed, dict) or parsed.get("v") != 1 or not parsed.get("at") or not parsed.get("requestId"):
+        raise RuntimeError("invalid page token")
+    return parsed
+
+def dict_by_id(rows):
+    return {str(row.get("id")): row for row in rows if isinstance(row, dict) and row.get("id") is not None}
+
+def keyed_map(value, key):
+    source = value.get(key) if isinstance(value, dict) else None
+    return source if isinstance(source, dict) else {}
+
+def runtime_health():
+    if RUNTIME_MODE == "host-docker":
+        proc = docker(["inspect", APP_CONTAINER, "sub2api-redis"])
+        if proc.returncode != 0:
+            return {"ok": False, "mode": RUNTIME_MODE, "error": text(proc.stderr)}
+        rows = []
+        for item in json.loads(proc.stdout.decode("utf-8")):
+            state = item.get("State") or {}
+            health = state.get("Health") or {}
+            rows.append({"name": str(item.get("Name") or "").lstrip("/"), "running": state.get("Running"), "status": health.get("Status") or state.get("Status")})
+        return {"ok": all(row.get("running") is True for row in rows), "mode": RUNTIME_MODE, "components": rows}
+    proc = run(["kubectl", "-n", NAMESPACE, "get", "pods", "-l", "app.kubernetes.io/name=sub2api", "-o", "json"])
+    if proc.returncode != 0:
+        return {"ok": False, "mode": RUNTIME_MODE, "error": text(proc.stderr)}
+    rows = []
+    for item in json.loads(proc.stdout.decode("utf-8")).get("items") or []:
+        status = item.get("status") or {}
+        containers = status.get("containerStatuses") or []
+        rows.append({"name": (item.get("metadata") or {}).get("name"), "running": status.get("phase") == "Running", "status": "ready" if containers and all(entry.get("ready") is True for entry in containers) else status.get("phase")})
+    return {"ok": bool(rows) and all(row.get("running") is True for row in rows), "mode": RUNTIME_MODE, "components": rows}
+
+def compact_log(row):
+    extra = row.get("extra") if isinstance(row.get("extra"), dict) else {}
+    allow = {}
+    for key in ("phase", "error_owner", "status_code", "duration_ms", "latency_ms", "time_to_first_token_ms", "upstream_status", "path", "stream"):
+        if key in extra:
+            allow[key] = extra.get(key)
+    return {
+        "id": row.get("id"),
+        "at": row.get("created_at"),
+        "level": row.get("level"),
+        "component": row.get("component"),
+        "message": redact_text(row.get("message")),
+        "requestId": row.get("request_id"),
+        "clientRequestId": row.get("client_request_id"),
+        "accountId": row.get("account_id"),
+        "extra": allow,
+    }
+
+def related_logs(request, logs):
+    ids = {str(request.get("request_id") or "")}
+    matched = []
+    changed = True
+    while changed:
+        changed = False
+        for row in logs:
+            request_id = str(row.get("request_id") or "")
+            client_id = str(row.get("client_request_id") or "")
+            if row in matched or (request_id not in ids and client_id not in ids):
+                continue
+            matched.append(row)
+            before = len(ids)
+            ids.update(value for value in (request_id, client_id) if value)
+            changed = changed or len(ids) != before
+    return matched
+
+def classify(request, logs, account, proxy, queued):
+    evidence = " ".join([str(request.get("phase") or ""), str(request.get("message") or ""), *[str(log.get("component") or "") + " " + str(log.get("message") or "") for log in logs]]).lower()
+    if any(marker in evidence for marker in ("client disconnect", "broken pipe", "context canceled", "client canceled")):
+        return "client-connection"
+    if any(marker in evidence for marker in ("proxy", "dial tcp", "dns", "connection refused", "network")) or (proxy and proxy.get("status") not in (None, "active")):
+        return "proxy-egress"
+    if any(marker in evidence for marker in ("upstream", "provider", "first token", "stream")) or str(request.get("phase") or "") == "upstream":
+        return "upstream-first-byte-stream"
+    if queued > 0:
+        return "sub2api-queue"
+    if request.get("kind") == "error" or (isinstance(request.get("status_code"), int) and request.get("status_code") >= 400):
+        return "sub2api-or-upstream-unknown"
+    return "completed"
+
+def main():
+    errors = []
+    admin_email, token = login()
+    user = exact_user(token)
+    if not isinstance(user, dict):
+        return {"ok": False, "mode": "feedback", "error": "user-not-found", "selector": USER_SELECTOR, "valuesPrinted": False}
+    user_id = user.get("id")
+    groups = safe_call(errors, "groups", lambda: items(api_data("GET", "/api/v1/admin/groups/all", token, label="groups")), [])
+    keys = safe_call(errors, "api-keys", lambda: page_all(token, f"/api/v1/admin/users/{user_id}/api-keys", {}, 100), [])
+    requests = safe_call(errors, "ops-requests", lambda: page_all(token, "/api/v1/admin/ops/requests", {"user_id": user_id, "time_range": WINDOW, "kind": "all", "sort": "created_at_desc"}, 100), [])
+    logs = safe_call(errors, "ops-system-logs", lambda: page_all(token, "/api/v1/admin/ops/system-logs", {"user_id": user_id, "time_range": WINDOW}, 200), [])
+    concurrency = safe_call(errors, "ops-concurrency", lambda: api_data("GET", "/api/v1/admin/ops/concurrency", token, label="concurrency"), {})
+    user_concurrency = safe_call(errors, "ops-user-concurrency", lambda: api_data("GET", "/api/v1/admin/ops/user-concurrency", token, label="user concurrency"), {})
+    availability = safe_call(errors, "ops-account-availability", lambda: api_data("GET", "/api/v1/admin/ops/account-availability", token, label="account availability"), {})
+    accounts = safe_call(errors, "accounts", lambda: page_all(token, "/api/v1/admin/accounts", {}, 100), [])
+    proxies = safe_call(errors, "proxies", lambda: page_all(token, "/api/v1/admin/proxies", {}, 100), [])
+    infrastructure = safe_call(errors, "infrastructure", runtime_health, {"ok": False, "mode": RUNTIME_MODE})
+
+    groups_by_id = dict_by_id(groups)
+    keys_by_id = dict_by_id(keys)
+    accounts_by_id = dict_by_id(accounts)
+    proxies_by_id = dict_by_id(proxies)
+    user_current_map = keyed_map(user_concurrency, "user")
+    account_concurrency_map = keyed_map(concurrency, "account")
+    account_availability_map = keyed_map(availability, "account")
+    current_user = user_current_map.get(str(user_id)) or {}
+    user_current = int(current_user.get("current_in_use") or user.get("current_concurrency") or 0)
+    user_waiting = int(current_user.get("waiting_in_queue") or 0)
+
+    compact_keys = []
+    for key in keys:
+        group = groups_by_id.get(str(key.get("group_id"))) or {}
+        compact_keys.append({
+            "id": key.get("id"), "name": key.get("name"), "status": key.get("status"),
+            "groupId": key.get("group_id"), "groupName": group.get("name"),
+            "currentConcurrency": key.get("current_concurrency"), "keyPrinted": False,
+        })
+
+    timeline = []
+    ordered = sorted([row for row in requests if isinstance(row, dict)], key=lambda row: iso_epoch(row.get("created_at")) or 0)
+    previous_epoch = None
+    for row in ordered:
+        matched_logs = related_logs(row, logs)
+        client_id = next((log.get("client_request_id") for log in matched_logs if log.get("client_request_id")), None)
+        account = accounts_by_id.get(str(row.get("account_id"))) or {}
+        proxy = proxies_by_id.get(str(account.get("proxy_id"))) or {}
+        account_load = account_concurrency_map.get(str(row.get("account_id"))) or {}
+        queued = int(account_load.get("waiting_in_queue") or 0)
+        epoch = iso_epoch(row.get("created_at"))
+        gap = round(epoch - previous_epoch, 1) if epoch is not None and previous_epoch is not None else None
+        previous_epoch = epoch if epoch is not None else previous_epoch
+        timeline.append({
+            "at": row.get("created_at"), "gapSeconds": gap, "kind": row.get("kind"),
+            "requestId": row.get("request_id"), "clientRequestId": client_id,
+            "model": row.get("model"), "statusCode": row.get("status_code"), "durationMs": row.get("duration_ms"),
+            "apiKeyId": row.get("api_key_id"), "apiKeyName": (keys_by_id.get(str(row.get("api_key_id"))) or {}).get("name"),
+            "accountId": row.get("account_id"), "accountName": account.get("name"),
+            "proxyId": account.get("proxy_id"), "proxyName": proxy.get("name"),
+            "attribution": classify(row, matched_logs, account, proxy, queued),
+            "evidence": {"systemLogCount": len(matched_logs), "accountStatus": account.get("status"), "accountSchedulable": account.get("schedulable"), "proxyStatus": proxy.get("status")},
+        })
+
+    all_timeline = timeline
+    max_gap = max([row.get("gapSeconds") for row in all_timeline if isinstance(row.get("gapSeconds"), (int, float))], default=None)
+    next_page_token = None
+    more_available = False
+    if REQUEST_SELECTOR:
+        selected_ids = {REQUEST_SELECTOR}
+        for row in all_timeline:
+            if row.get("requestId") == REQUEST_SELECTOR or row.get("clientRequestId") == REQUEST_SELECTOR:
+                selected_ids.update(value for value in (row.get("requestId"), row.get("clientRequestId")) if isinstance(value, str) and value)
+        timeline = [row for row in all_timeline if any(isinstance(value, str) and value in selected_ids for value in (row.get("requestId"), row.get("clientRequestId")))]
+        detail_logs = [compact_log(row) for row in logs if any(isinstance(value, str) and value in selected_ids for value in (row.get("request_id"), row.get("client_request_id")))]
+    else:
+        detail_logs = []
+        ordered_desc = list(reversed(all_timeline))
+        cursor = decode_page_token(PAGE_TOKEN)
+        start = 0
+        if cursor:
+            start = next((index + 1 for index, row in enumerate(ordered_desc) if row.get("at") == cursor.get("at") and row.get("requestId") == cursor.get("requestId")), -1)
+            if start < 0:
+                raise RuntimeError("page token cursor is no longer present in the selected window")
+        timeline = ordered_desc[start:start + PAGE_SIZE]
+        more_available = start + len(timeline) < len(ordered_desc)
+        if more_available and timeline:
+            next_page_token = encode_page_token(timeline[-1])
+    if not requests and user_waiting > 0:
+        conclusion = "sub2api-queue"
+    elif not requests and user_current > 0:
+        conclusion = "in-flight"
+    elif not requests:
+        conclusion = "client-before-submit-or-no-inbound-evidence"
+    elif timeline:
+        conclusion = timeline[0].get("attribution")
+    else:
+        conclusion = "request-selector-not-found"
+
+    related_account_ids = {str(row.get("accountId")) for row in timeline if row.get("accountId") is not None}
+    account_evidence = []
+    for account_id in sorted(related_account_ids):
+        account = accounts_by_id.get(account_id) or {}
+        proxy = proxies_by_id.get(str(account.get("proxy_id"))) or {}
+        load = account_concurrency_map.get(account_id) or {}
+        available = account_availability_map.get(account_id) or {}
+        account_evidence.append({
+            "id": account.get("id"), "name": account.get("name"), "status": account.get("status"), "schedulable": account.get("schedulable"),
+            "currentInUse": load.get("current_in_use"), "waitingInQueue": load.get("waiting_in_queue"), "maxCapacity": load.get("max_capacity"),
+            "available": available.get("is_available"), "rateLimited": available.get("is_rate_limited"), "overloaded": available.get("is_overloaded"),
+            "proxyId": account.get("proxy_id"), "proxyName": proxy.get("name"), "proxyStatus": proxy.get("status"),
+        })
+
+    return {
+        "ok": True,
+        "mode": "feedback",
+        "selector": {"user": USER_SELECTOR, "requestId": REQUEST_SELECTOR, "pageToken": PAGE_TOKEN, "window": WINDOW},
+        "user": {"id": user_id, "email": user.get("email"), "username": user.get("username"), "status": user.get("status"), "maxConcurrency": user.get("concurrency"), "currentInUse": user_current, "waitingInQueue": user_waiting},
+        "apiKeys": compact_keys,
+        "lifecycle": {"notInbound": not requests and user_current == 0 and user_waiting == 0, "queued": user_waiting, "inFlight": user_current, "completed": len(requests)},
+        "timeline": timeline,
+        "timelineSummary": {"returned": len(timeline), "total": len(all_timeline), "moreAvailable": more_available, "nextPageToken": next_page_token, "pageSize": PAGE_SIZE, "systemLogCount": len(logs), "maxGapSeconds": max_gap},
+        "requestDetail": {"logs": detail_logs, "logCount": len(detail_logs)} if REQUEST_SELECTOR else None,
+        "accounts": account_evidence,
+        "infrastructure": infrastructure,
+        "conclusion": {"attribution": conclusion, "boundary": "只读证据归因；无入站记录不等于客户端未调用，运行中请求仅由实时并发佐证。"},
+        "toolCallReduction": {"manualInvestigationCalls": 9, "feedbackCliCalls": 1, "reducedCalls": 8, "reductionPercent": 88.9, "basis": "issue-2000: user, requests, system logs, two request-id traces, account, proxy, process/port, journal"},
+        "sources": ["admin users", "admin user api-keys", "ops requests", "ops system-logs", "ops concurrency", "ops user-concurrency", "ops account-availability", "admin accounts", "admin proxies", "runtime health"],
+        "errors": errors,
+        "admin": {"email": admin_email, "tokenPrinted": False},
+        "valuesPrinted": False,
+    }
+
+try:
+    print(json.dumps(main(), ensure_ascii=False))
+except Exception as exc:
+    print(json.dumps({"ok": False, "mode": "feedback", "error": redact_text(str(exc)), "valuesPrinted": False}, ensure_ascii=False))
+    raise
+PY
+`;
+}
+
+function record(value: unknown): Record<string, unknown> {
+  return typeof value === "object" && value !== null && !Array.isArray(value) ? value as Record<string, unknown> : {};
+}
+
+function records(value: unknown): Record<string, unknown>[] {
+  return Array.isArray(value) ? value.filter((item): item is Record<string, unknown> => typeof item === "object" && item !== null && !Array.isArray(item)) : [];
+}
+
+function renderFeedbackReport(report: Record<string, unknown> | null, options: FeedbackOptions, remote: Record<string, unknown>): string {
+  if (report === null) {
+    return [
+      `SUB2API FEEDBACK user=${options.user} unavailable`,
+      `remote_exit=${remote.exitCode ?? "?"} stdout_bytes=${remote.stdoutBytes ?? "?"} stderr_bytes=${remote.stderrBytes ?? "?"}`,
+      textValue(remote.stderrTail ?? remote.stdoutTail),
+    ].join("\n");
+  }
+  if (report.ok !== true) {
+    return [
+      `SUB2API FEEDBACK user=${options.user} ok=false`,
+      `error=${textValue(report.error)}`,
+      "JSON: add --json for the structured failure envelope.",
+    ].join("\n");
+  }
+  const user = record(report.user);
+  const lifecycle = record(report.lifecycle);
+  const summary = record(report.timelineSummary);
+  const conclusion = record(report.conclusion);
+  const reduction = record(report.toolCallReduction);
+  const timeline = records(report.timeline);
+  const accounts = records(report.accounts);
+  const apiKeys = records(report.apiKeys);
+  const detail = record(report.requestDetail);
+  const detailLogs = records(detail.logs);
+  const infrastructure = record(report.infrastructure);
+  const components = records(infrastructure.components);
+  const errors = records(report.errors);
+  const lines: string[] = [];
+  lines.push(`SUB2API FEEDBACK user=${textValue(user.email)}#${textValue(user.id)} window=${options.window} ok=true`);
+  lines.push(`CONCLUSION attribution=${textValue(conclusion.attribution)} boundary=${textValue(conclusion.boundary)}`);
+  lines.push(`LIFECYCLE not_inbound=${textValue(lifecycle.notInbound)} queued=${textValue(lifecycle.queued)} in_flight=${textValue(lifecycle.inFlight)} completed=${textValue(lifecycle.completed)} max_gap_s=${textValue(summary.maxGapSeconds)}`);
+  lines.push(`TOOLS manual=${textValue(reduction.manualInvestigationCalls)} cli=${textValue(reduction.feedbackCliCalls)} reduced=${textValue(reduction.reducedCalls)} reduction=${textValue(reduction.reductionPercent)}%`);
+  if (apiKeys.length > 0) {
+    lines.push("");
+    lines.push("API KEYS");
+    lines.push(renderTable([
+      ["ID", "NAME", "GROUP", "STATUS", "CURRENT"],
+      ...apiKeys.map((key) => [textValue(key.id), shorten(textValue(key.name), 30), shorten(textValue(key.groupName), 24), textValue(key.status), textValue(key.currentConcurrency)]),
+    ]));
+  }
+  lines.push("");
+  lines.push(`REQUESTS returned=${textValue(summary.returned)} total=${textValue(summary.total)} moreAvailable=${textValue(summary.moreAvailable)} system_logs=${textValue(summary.systemLogCount)}`);
+  if (timeline.length === 0) {
+    lines.push("No matching completed request records in the selected window.");
+  } else {
+    lines.push(renderTable([
+      ["#", "AT", "GAP_S", "STATUS", "DURATION", "MODEL", "ACCOUNT", "ATTRIBUTION"],
+      ...timeline.map((item, index) => [
+        String(index + 1), shortIso(item.at), textValue(item.gapSeconds), textValue(item.statusCode ?? item.kind), textValue(item.durationMs),
+        shorten(textValue(item.model), 18), shorten(`${textValue(item.accountName)}#${textValue(item.accountId)}`, 28), textValue(item.attribution),
+      ]),
+    ]));
+    lines.push("");
+    lines.push("REQUEST ID INDEX");
+    lines.push(renderTable([
+      ["#", "REQUEST_ID", "CLIENT_ID"],
+      ...timeline.map((item, index) => [String(index + 1), textValue(item.requestId), textValue(item.clientRequestId)]),
+    ]));
+  }
+  if (accounts.length > 0 || components.length > 0) {
+    lines.push("");
+    lines.push("ACCOUNT / INFRA");
+    lines.push(renderTable([
+      ["ACCOUNT", "STATUS", "SCHED", "IN_USE", "QUEUE", "CAP", "AVAILABLE", "PROXY", "P_STATUS"],
+      ...accounts.map((item) => [
+        shorten(`${textValue(item.name)}#${textValue(item.id)}`, 32), textValue(item.status), textValue(item.schedulable),
+        textValue(item.currentInUse), textValue(item.waitingInQueue), textValue(item.maxCapacity), textValue(item.available),
+        shorten(`${textValue(item.proxyName)}#${textValue(item.proxyId)}`, 24), textValue(item.proxyStatus),
+      ]),
+      ...components.map((item) => [shorten(textValue(item.name), 32), textValue(item.status), "-", "-", "-", "-", textValue(item.running), "runtime", textValue(infrastructure.mode)]),
+    ]));
+  }
+  if (options.requestId !== null) {
+    lines.push("");
+    lines.push(`REQUEST DETAIL id=${options.requestId} logs=${textValue(detail.logCount)}`);
+    if (detailLogs.length > 0) {
+      lines.push(renderTable([
+        ["LOG_ID", "AT", "LEVEL", "COMPONENT", "REQUEST_ID", "CLIENT_ID", "MESSAGE"],
+        ...detailLogs.map((item) => [
+          textValue(item.id), shortIso(item.at), textValue(item.level), shorten(textValue(item.component), 22),
+          shorten(textValue(item.requestId), 18), shorten(textValue(item.clientRequestId), 18), shorten(textValue(item.message), 64),
+        ]),
+      ]));
+    }
+  }
+  if (errors.length > 0) {
+    lines.push("");
+    lines.push("PARTIAL EVIDENCE");
+    lines.push(renderTable([["SOURCE", "ERROR"], ...errors.map((item) => [textValue(item.source), shorten(textValue(item.error), 90)])]));
+  }
+  lines.push("");
+  if (summary.moreAvailable === true && typeof summary.nextPageToken === "string") {
+    lines.push(`NEXT_PAGE_TOKEN ${summary.nextPageToken}`);
+    lines.push(`Next: bun scripts/cli.ts platform-infra sub2api codex-pool feedback --target ${options.targetId} --user ${options.user} --window ${options.window} --page-token ${summary.nextPageToken}`);
+  }
+  lines.push("Disclosure: rerun with --request-id <client-or-internal-id> for correlated indexed logs.");
+  lines.push("JSON: add --json for the complete redacted report.");
+  return lines.join("\n");
+}

scripts/src/platform-infra-sub2api-codex/index.ts

@@ -13,3 +13,4 @@ export * from "./remote-scripts";
 export * from "./accounts";
 export * from "./remote-python-sync";
 export * from "./remote";
+export * from "./feedback";

scripts/src/platform-infra-sub2api-codex/options.ts

@@ -21,8 +21,9 @@ import {
 import { parseEnvFile, readTextFile, redactRepoPath, requiredEnvValue } from "../secrets";
 import { runSshCommandCapture, type SshCaptureResult } from "../ssh";
 
-import type { ConfirmOptions, DisclosureOptions, SentinelImageOptions, SentinelProbeOptions, SentinelReportOptions, SyncOptions, TraceOptions } from "./types";
+import type { ConfirmOptions, DisclosureOptions, FeedbackOptions, SentinelImageOptions, SentinelProbeOptions, SentinelReportOptions, SyncOptions, TraceOptions } from "./types";
 import { codexPoolCleanupProbes, codexPoolConfigureLocal, codexPoolExpose, codexPoolPlan, codexPoolSentinelImage, codexPoolSentinelProbe, codexPoolSentinelReport, codexPoolSync, codexPoolTrace, codexPoolValidate } from "./actions";
+import { codexPoolFeedback } from "./feedback";
 import { renderCodexPoolPlan } from "./render";
 import { codexPoolRuntime } from "./runtime";
 import { defaultCodexPoolRuntimeTargetId } from "./runtime-target";
@@ -39,6 +40,7 @@ export async function runCodexPoolCommand(config: UniDeskConfig, args: string[])
   if (action === "validate") return await codexPoolValidate(config, parseDisclosureOptions(args.slice(1)));
   if (action === "runtime") return await codexPoolRuntime(config, args.slice(1));
   if (action === "trace") return await codexPoolTrace(config, parseTraceOptions(args.slice(1)));
+  if (action === "feedback") return await codexPoolFeedback(config, parseFeedbackOptions(args.slice(1)));
   if (action === "sentinel-image") return await codexPoolSentinelImage(config, parseSentinelImageOptions(args.slice(1)));
   if (action === "sentinel-probe") return await codexPoolSentinelProbe(config, parseSentinelProbeOptions(args.slice(1)));
   if (action === "sentinel-report") return await codexPoolSentinelReport(config, parseSentinelReportOptions(args.slice(1)));
@@ -181,6 +183,7 @@ export function parseSentinelReportOptions(args: string[]): SentinelReportOption
 
 export function parseTraceOptions(args: string[]): TraceOptions {
   let requestId: string | null = null;
+  let pageToken: string | null = null;
   let since = "24h";
   let tail = 20_000;
   let contextSeconds = 300;
@@ -258,6 +261,64 @@ export function parseTraceOptions(args: string[]): TraceOptions {
   return { ...disclosure, requestId, since, tail, contextSeconds, showLines };
 }
 
+export function parseFeedbackOptions(args: string[]): FeedbackOptions {
+  let user: string | null = null;
+  let window: FeedbackOptions["window"] = "30m";
+  let requestId: string | null = null;
+  let pageToken: string | null = null;
+  let json = false;
+  const targetArgs: string[] = [];
+  const windows = new Set<FeedbackOptions["window"]>(["5m", "30m", "1h", "6h", "24h", "7d", "30d"]);
+  for (let index = 0; index < args.length; index += 1) {
+    const arg = args[index]!;
+    if (arg === "--json") {
+      json = true;
+      continue;
+    }
+    if (arg === "--target" || arg.startsWith("--target=")) {
+      targetArgs.push(arg);
+      if (arg === "--target") {
+        const value = args[index + 1];
+        if (value === undefined || value.startsWith("--")) throw new Error("--target requires a value");
+        targetArgs.push(value);
+        index += 1;
+      }
+      continue;
+    }
+    const readValue = (option: string): string => {
+      const value = args[index + 1];
+      if (value === undefined || value.startsWith("--")) throw new Error(`${option} requires a value`);
+      index += 1;
+      return value.trim();
+    };
+    if (arg === "--user") user = readValue(arg);
+    else if (arg.startsWith("--user=")) user = arg.slice("--user=".length).trim();
+    else if (arg === "--window") {
+      const value = readValue(arg) as FeedbackOptions["window"];
+      if (!windows.has(value)) throw new Error("--window must be one of 5m, 30m, 1h, 6h, 24h, 7d, 30d");
+      window = value;
+    } else if (arg.startsWith("--window=")) {
+      const value = arg.slice("--window=".length) as FeedbackOptions["window"];
+      if (!windows.has(value)) throw new Error("--window must be one of 5m, 30m, 1h, 6h, 24h, 7d, 30d");
+      window = value;
+    } else if (arg === "--request-id") requestId = readValue(arg);
+    else if (arg.startsWith("--request-id=")) requestId = arg.slice("--request-id=".length).trim();
+    else if (arg === "--page-token") pageToken = readValue(arg);
+    else if (arg.startsWith("--page-token=")) pageToken = arg.slice("--page-token=".length).trim();
+    else throw new Error(`unsupported option: ${arg}`);
+  }
+  if (user === null || user.length === 0) throw new Error("feedback requires --user <email-or-id>");
+  if (user.length > 320 || /[\r\n]/u.test(user)) throw new Error("--user must be a single email or positive user id");
+  if (requestId !== null && (requestId.length === 0 || requestId.length > 256 || /[\r\n]/u.test(requestId))) {
+    throw new Error("--request-id must be a non-empty stable request id up to 256 characters");
+  }
+  if (pageToken !== null && (pageToken.length === 0 || pageToken.length > 1024 || !/^[A-Za-z0-9_-]+$/u.test(pageToken))) {
+    throw new Error("--page-token must be a URL-safe cursor token");
+  }
+  if (requestId !== null && pageToken !== null) throw new Error("--request-id and --page-token cannot be combined");
+  return { user, window, requestId, pageToken, json, targetId: parseTargetId(targetArgs) };
+}
+
 export function parseKubectlDuration(raw: string): string {
   const value = raw.trim();
   if (!/^[1-9][0-9]*(?:s|m|h)$/u.test(value)) throw new Error("--since must be a kubectl duration such as 24h, 90m, or 300s");

scripts/src/platform-infra-sub2api-codex/types.ts

@@ -71,6 +71,15 @@ export interface TraceOptions extends DisclosureOptions {
   showLines: boolean;
 }
 
+export interface FeedbackOptions {
+  user: string;
+  window: "5m" | "30m" | "1h" | "6h" | "24h" | "7d" | "30d";
+  requestId: string | null;
+  pageToken: string | null;
+  json: boolean;
+  targetId: string;
+}
+
 export interface SentinelImageOptions extends DisclosureOptions {
   action: "status" | "build";
   confirm: boolean;
@@ -341,8 +350,8 @@ export function codexPoolHelp(): unknown {
   const pool = readCodexPoolConfig();
   const runtimeTarget = codexPoolRuntimeTarget();
   return {
-    command: "platform-infra sub2api codex-pool plan|sync|validate|runtime|trace|sentinel-image|sentinel-probe|sentinel-report|cleanup-probes|expose|configure-local",
-    output: "json, except trace and sentinel-report default to low-noise text tables",
+    command: "platform-infra sub2api codex-pool plan|sync|validate|runtime|trace|feedback|sentinel-image|sentinel-probe|sentinel-report|cleanup-probes|expose|configure-local",
+    output: "json, except trace, feedback, and sentinel-report default to low-noise text tables",
     usage: [
       "bun scripts/cli.ts platform-infra sub2api codex-pool plan",
       "bun scripts/cli.ts platform-infra sub2api codex-pool plan --target D601",
@@ -354,6 +363,7 @@ export function codexPoolHelp(): unknown {
       "bun scripts/cli.ts platform-infra sub2api codex-pool runtime apply --account <name-or-id> --template <id> [--target PK01] [--confirm]",
       "bun scripts/cli.ts platform-infra sub2api codex-pool runtime delete --account <name-or-id> --kind temp-unschedulable [--target PK01] [--confirm]",
       "bun scripts/cli.ts platform-infra sub2api codex-pool trace [--target D601] --request-id <requestId> [--since 24h|--tail 20000|--context-seconds 300|--show-lines|--raw]",
+      "bun scripts/cli.ts platform-infra sub2api codex-pool feedback --user <email-or-id> [--window 30m] [--page-token <token>|--request-id <client-or-internal-id>] [--target PK01] [--json]",
       "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-image status [--target D601]",
       "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-image build [--target D601] --confirm",
       "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-probe [--target D601] --account unidesk-codex-hy --confirm",

scripts/src/platform-infra/entry.ts

@@ -421,6 +421,7 @@ export function platformInfraHelp(): unknown {
       "bun scripts/cli.ts platform-infra sub2api codex-pool sync --confirm",
       "bun scripts/cli.ts platform-infra sub2api codex-pool validate",
       "bun scripts/cli.ts platform-infra sub2api codex-pool trace --request-id <requestId>",
+      "bun scripts/cli.ts platform-infra sub2api codex-pool feedback --user <email-or-id> [--window 30m] [--page-token <token>|--request-id <client-or-internal-id>] [--json]",
       "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-image status",
       "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-probe --account unidesk-codex-hy --confirm",
       "bun scripts/cli.ts platform-infra sub2rank plan [--target NC01]",
@@ -503,6 +504,7 @@ export function platformInfraHelp(): unknown {
         "bun scripts/cli.ts platform-infra sub2api codex-pool sync --confirm",
         "bun scripts/cli.ts platform-infra sub2api codex-pool validate",
         "bun scripts/cli.ts platform-infra sub2api codex-pool trace --request-id <requestId>",
+        "bun scripts/cli.ts platform-infra sub2api codex-pool feedback --user <email-or-id> [--window 30m] [--page-token <token>|--request-id <client-or-internal-id>] [--json]",
         "bun scripts/cli.ts platform-infra sub2api codex-pool sentinel-image status",
       ],
       module: "scripts/src/platform-infra-sub2api-codex.ts",