メニュー
Scrape training metrics from Tencent AI Arena (tencentarena.com) monitoring dashboard. Covers JWT auth, x-kaiwu-auth signature, API endpoints, and metric extraction.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
把静态 HTML 站点部署到服务器对外访问的标准 SOP。覆盖 Caddy(Docker,:80/:443) 与 nginx(宿主,独立端口) 两条路径、端口约定、80 端口冲突坑、部署目录、reload 与 curl 验证。任何"部署/上线/开放访问一个 HTML 页面或静态站点"的请求都必须先读这里，不要自己瞎配。
Diagnose RL agent behavioral issues by analyzing reward structure. Use when the agent exhibits unwanted behavior (random skills, wandering, avoiding combat, etc.) to identify root causes in the reward function.
Author/validate/export Google's DESIGN.md token spec files.
Configure, troubleshoot, and operate the Feishu (Lark) gateway integration — group chats, @mention gating, permissions, message routing, and diagnostics.
Write and maintain Kaiwu_2026 答辩材料 (defense/presentation docs). Covers technical writing for innovation points, updating the main summary doc, and git workflow for doc commits.
Use when working with the Kaiwu platform monitoring system — fetching training metrics, plotting charts, running anomaly checks, or updating monitor config. Covers the monitoring/ directory: fetch.py, check.py, plot.py, config.yaml.
| name | arena-monitor |
| description | Scrape training metrics from Tencent AI Arena (tencentarena.com) monitoring dashboard. Covers JWT auth, x-kaiwu-auth signature, API endpoints, and metric extraction. |
| version | 1.0.0 |
| author | QoobeeHermes |
| platforms | ["linux","macos","windows"] |
| metadata | {"hermes":{"tags":["kaiwu","monitoring","api","metrics"],"related_skills":["rl-advisor","repo-sync"]}} |
从腾讯开悟平台 (tencentarena.com) 抓取训练监控数据。
JWT token 从浏览器 cookie kaiwu-token 获取，同时作为 Authorization: Bearer 头。
JWT payload 包含：
custom: user_id（如 15850）exp: 过期时间（通常 7 天）iss: "kaiwu"每个 API 请求需要两个自定义头：
x-kaiwu-ts: <当前 Unix 时间戳（秒）>
x-kaiwu-auth: DJB2_HASH(timestamp + token_last_32_chars + url_last_path_segment)
DJB2 hash 实现（JavaScript 源码）：
function k9(t) {
return function(t, e) {
for (var r = 5381, n = 0, o = e.length; n < o; ++n)
r += (r << 5) + e.charCodeAt(n);
return 2147483647 & r
}(0, t)
}
Python 实现：
def kaiwu_auth_hash(timestamp: int, token: str, url_path: str) -> int:
last32 = token[-32:]
last_segment = url_path.rstrip('/').split('/')[-1]
data = f"{timestamp}{last32}{last_segment}"
r = 5381
for c in data:
r = (r + (r << 5) + ord(c)) & 0xFFFFFFFF
return r & 2147483647
Cookie: select_lang=zh; kaiwu-token=<JWT>
Authorization: Bearer <JWT>
x-kaiwu-auth: <hash>
x-kaiwu-ts: <timestamp>
Referer: https://tencentarena.com/p/v5/exp/monitor
基础路径：https://tencentarena.com/api/v5/Competition/
| 端点 | 方法 | 说明 |
|---|---|---|
GetTrainTask | POST | 获取训练任务信息 |
GetTrainLog | POST | 获取训练日志 |
GetTrainMetric | POST | 获取指标数据（PromQL 查询） |
GetTrainMetricRange | POST | 获取指标时间范围 |
ListTrainTask | POST | 列出训练任务 |
ListTrainAiModel | POST | 列出训练模型 |
GetExperiment | POST | 获取实验信息 |
{
"domain": {"type": "competition_stage", "id": 444},
"train_task_id": 183628,
"competition_team_id": 7203,
"experiment_id": 11425,
"start_time": {"timestamp": "2026-05-27T16:00:00Z"},
"end_time": {"timestamp": "2026-05-28T02:00:00Z"},
"queries": [
{"expr": "round(avg(kaiwu_win_rate{model_id=\"selfplay\"}) by (model_id), 0.01)"}
]
}
kaiwu_train_global_step — 全局训练步数kaiwu_actor_predict_succ_cnt — Actor 预测成功数kaiwu_sample_production_and_consumption_ratio — 样本生产消费比kaiwu_episode_cnt — 对局数kaiwu_actor_load_last_model_succ_cnt — Actor 加载最新模型成功数kaiwu_sample_receive_cnt — 样本接收数kaiwu_win_rate — 胜率（selfplay 对局）kaiwu_self_tower_hp / kaiwu_enemy_tower_hp — 塔血量kaiwu_frame — 对局帧数kaiwu_kill / kaiwu_death — 击杀/死亡数kaiwu_money_per_frame — 每帧经济kaiwu_hurt_to_hero / kaiwu_hurt_by_hero — 每帧伤害kaiwu_direnjie_vs_luban_win_rate、kaiwu_luban_vs_direnjie_win_ratekaiwu_eval_luban_win_rate、kaiwu_eval_direnjie_win_ratekaiwu_luban_summoner_win_rate、kaiwu_direnjie_summoner_win_ratekaiwu_total_loss、kaiwu_value_loss、kaiwu_policy_loss、kaiwu_entropy_losskaiwu_value_loss_hp、kaiwu_value_loss_tower、kaiwu_value_loss_econ、kaiwu_value_loss_combatkaiwu_learning_rate、kaiwu_entropy_beta、kaiwu_grad_normkaiwu_sample_batch_size、kaiwu_train_stepkaiwu_reward（平台累积回报）、kaiwu_reward_sum（对局总回报）kaiwu_reward_hp、kaiwu_reward_tower、kaiwu_reward_econ、kaiwu_reward_combatkaiwu_episode_frame、kaiwu_episode_win_value监控页面是 Next.js RSC (React Server Components) 应用。页面 HTML 中嵌入了：
提取方法：用 curl 带认证头请求页面 URL，然后从 self.__next_f.push([1, "..."]) payload 中解析 JSON。
| 参数 | 值 |
|---|---|
| task_id | 189696 |
| task_uuid | 57738cce-5bfb-4183-8ab2-37b57bb407cb |
| team_id | 7203 |
| domain_id | 444 |
| exp_id | 11425 |
| algorithm | PPO |
| training_mode | Distributed |
| status | running |
model_id=\"selfplay\" 需要反斜杠转义