← 返回 Skills 市场
79
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install ak-data-daily-timeout-report
功能描述
Unified daily timeout entry for AK Data single job type: task timeout + crawler timeout + comparison + drill-down examples with log request.
使用说明 (SKILL.md)
Skill: ak-data-daily-timeout-report
可迁移入口脚本:
./run.sh
- 单天查询:
./run.sh day ...- 趋势查询:
./run.sh trend ...- 迁移说明:
./MIGRATION.md- 模板说明:
./TEMPLATE.md- 短命令包装:项目根目录
bin/timeout- 打包脚本:
scripts/data/package_ak-data_daily_timeout_skill.py
统一入口:用于查询 AK Data 单任务类型 在某一天的:
- 主任务超时(task timeout)
- 爬虫超时(crawler timeout)
- 两种口径对比结论
- 支持继续追问具体任务例子(自动补齐 log
ext_ssn + log_state + request)
这是入口 Skill,不直接替代底层统计逻辑。
- 主任务超时底层能力:
ak-data-task-timeout-created-report- 爬虫超时底层能力:
ak-data-depart-timeout-log-report本 Skill 负责:统一调度、统一输出、统一追问接口。
快速使用
1) 单天联合超时
./run.sh day --date-bjt 2026-04-05 --job-type AmazonListingJob
2) 单天 + 样例追问
./run.sh day \
--date-bjt 2026-04-05 \
--job-type AmazonListingJob \
--example-scope crawler_timeout \
--example-type overtime_total \
--example-market au \
--example-limit 5
3) 区间趋势
./run.sh trend \
--start-date-bjt 2026-04-01 \
--end-date-bjt 2026-04-05 \
--job-type AmazonListingJob
适用场景
当 LeeHoo 的问题是以下类型时,优先使用本 Skill:
- “查 4.5 超时情况”
- “查 4.5 主任务超时和爬虫超时对比”
- “查某天 AmazonListingJob 的超时情况”
- “给我几个超时例子,带 request”
- “继续看 4.5 的 nl / jp / au 样例”
- “导出未Deliver / 未Arrival / WAITING 样例”
口径边界(强制)
1) 主任务超时(task timeout)
- 时间窗口:
created_at落在dateBjt对应自然日 - 正式超时判定基线:
break_at - 不得把主任务超时偷换成
depart_at口径 - 不得把主任务超时简化成固定
created_at + 12h口径后对外宣称“正式口径”
2) 爬虫超时(crawler timeout)
- 时间窗口:
depart_at落在dateBjt对应自然日 - 超时判定:
- 已到达:
arrival_at - depart_at > timeoutHours - 未到达:
now_utc - depart_at > timeoutHours
- 已到达:
- 若查询日是当天,可启用
cutoffHours做成熟母本截断
3) 追问具体任务例子
例子查询必须能区分:
scopeType = task_timeoutscopeType = crawler_timeout
不得混口径返回。
输入契约
{
"tenantId": "AK Data",
"jobType": "AmazonListingJob",
"dateBjt": "2026-04-05",
"includeTaskTimeout": true,
"includeCrawlerTimeout": true,
"samplePerState": 3,
"needMarketSplit": true,
"needLogEnrich": true,
"cutoffHours": 0,
"exampleQuery": {
"scopeType": "crawler_timeout",
"sampleType": "overtime_total",
"market": "au",
"status": "PENDING",
"limit": 5,
"includeRequest": true
}
}
字段说明
includeTaskTimeout:是否生成主任务超时部分includeCrawlerTimeout:是否生成爬虫超时部分samplePerState:底层脚本每类抽样条数needLogEnrich:是否自动补齐 log 字段cutoffHours:仅爬虫口径支持;用于当日成熟母本截断exampleQuery:可选,用于继续追问例子
标准执行流程(强制)
A. 每日总览查询
- 如果
includeTaskTimeout=true- 调用
run_created_timeout_report.py - 读取输出 JSON
- 调用
- 如果
includeCrawlerTimeout=true- 调用
run_depart_timeout_report.py - 读取输出 JSON
- 调用
- 生成 unified report:
- 总览
- market 对比
- 主任务 vs 爬虫差值
- 诊断结论
B. 对比诊断规则
至少输出:
- 主任务超时率
- 爬虫超时率
- gap(百分点)
- top market 重合情况
- 结论
建议判定规则
- 若两者超时率差值
\x3C 1 pct- 判定:基本贴脸一致,问题主要由 crawler 链路传导
- 若主任务明显高于爬虫
> 2 pct- 判定:deliver / 状态收口尾段可能存在额外拖后腿
- 若 top risk market 高度重合
- 判定:风险源一致
- 若主任务新增 crawler 不明显的 market
- 判定:后链路异常需单独下钻
C. 追问例子(drill-down)
当用户继续问“给几个例子”时:
- 优先复用对应 JSON 中的
samples - 若用户指定过滤维度(market/status/sampleType),则从标准化样例集合中过滤
- 若默认样例不足,再调用底层脚本增强抽样能力或补充专用导出脚本
默认要补齐的字段:
req_ssnsource_tablelog_tablemarketstatuscreated_at / depart_at / arrival_at / deliver_attask_hours / crawl_hoursext_ssnlog_staterequestrequest_truncated
输出契约
{
"scope": {
"tenantId": "AK Data",
"jobType": "AmazonListingJob",
"dateBjt": "2026-04-05"
},
"taskTimeout": {
"baseline": "break_at",
"generatedAtUtc": "2026-04-06 08:00:00",
"total": {},
"markets": [],
"samples": {}
},
"crawlerTimeout": {
"baseline": "depart_at",
"generatedAtUtc": "2026-04-06 08:00:00",
"total": {},
"markets": [],
"samples": {}
},
"comparison": {
"taskTimeoutRate": 0.087,
"crawlerTimeoutRate": 0.0854,
"gapPct": 0.16,
"topOverlapMarkets": ["nl", "jp", "pl"],
"diagnosis": "主任务超时与爬虫超时基本贴脸一致,说明问题主要由 crawler 链路传导"
},
"exampleQueryResult": {
"scopeType": "crawler_timeout",
"sampleType": "overtime_total",
"market": "au",
"status": "PENDING",
"limit": 5,
"rows": []
}
}
输出文案要求(对话)
对 LeeHoo 汇报时,建议固定结构:
- 执行范围
- 总览指标
- 主任务 vs 爬虫差值
- Top market / 高风险 market
- 诊断结论
- 如用户要求,补具体样例(带 request)
- 最后给出文件路径
风险与降级
- 若某一底层报告失败:
- 返回另一部分成功结果
- 明确标记
partial=true
- 若 log 缺失:
- 保留主记录
log_found=false
- 若
request非法 JSON:- 返回原文字符串
request_parse_error=true
- 若 sample 不足:
- 明确提示“默认样例不足,当前仅返回已有样例”
结论型经验(已沉淀)
- LeeHoo 对该类问题的真实诉求不是单一口径,而是:
- 某天总体超时怎样
- 主任务超时与爬虫超时是否贴脸一致
- 是否属于 crawler 传导,而不是 deliver 尾段独立炸锅
- 继续追问具体任务例子,并看 request 字段
- 因此统一入口 Skill 比两个分散 Skill 更适合作为对话层默认路由。
安全使用建议
Do not install or run this skill as-is in a production environment. Key points to check before using: 1) The bundle is incomplete — ask the publisher for the referenced scripts (scripts/data/run_*.py) and the packaging bundle; the SKILL.md/MIGRATION.md assume those exist. 2) Confirm exactly which environment variables and configuration files are required (the MIGRATION.md lists ANKER_JOB_DB_* variables and a configs JSON) and ensure the skill metadata is updated to declare them. 3) Inspect the missing Python scripts for any network I/O or exfiltration code (outbound requests, sockets, uploads) before providing DB credentials. 4) If you must run it, use a dedicated read-only DB account limited to the minimal tables and fields needed, run in an isolated environment, and avoid including production secrets in a shared .env. 5) If the skill will return 'request' payloads, ensure you understand and redact any sensitive data. If the publisher/source is unknown or cannot provide the full scripts and a security review, treat this package as unsafe to run.
功能分析
Type: OpenClaw Skill
Name: ak-data-daily-timeout-report
Version: 1.0.0
The skill bundle 'ak-data-daily-timeout-report' is a legitimate data reporting tool designed to analyze job and crawler timeouts. It consists of an entry-point shell script (run.sh), detailed agent instructions (SKILL.md), and migration/template documentation. The skill functions by executing local Python scripts to query a database for performance metrics and logs. While it requires database credentials and executes external scripts, these capabilities are strictly aligned with its stated purpose of generating business reports. No evidence of malicious intent, data exfiltration, or prompt injection attacks was found.
能力评估
Purpose & Capability
The SKILL.md and MIGRATION.md describe a unified entry that orchestrates several Python reporting scripts and reads job/log data, which is coherent with the name. However, the actual skill bundle does NOT include the referenced Python scripts (scripts/data/run_*.py) and the registry metadata declares no environment variables or credentials — yet MIGRATION.md explicitly requires DB credentials and config. The declared requirements do not match the actual capabilities/dependencies.
Instruction Scope
Runtime instructions (run.sh and SKILL.md) tell the agent to execute Python reporting scripts, read produced JSON, and optionally return raw 'request' payloads from logs. Those actions require access to project files, DB credentials (.env), and underlying scripts that are missing from the bundle. The SKILL.md also instructs including potentially sensitive fields (request, req_ssn) in outputs; that expands the data surface.
Install Mechanism
There is no install spec (instruction-only + run.sh), which is low risk in itself. But MIGRATION.md says the pack needs to be bundled and notes Python 3.10+ and pip dependency 'pymysql'. Those runtime deps are not declared in metadata. The run.sh execs python3 scripts located outside the skill folder (ROOT_DIR/scripts/...), so the packaged skill as provided is unusable without copying extra files.
Credentials
Registry metadata lists no required env vars, but MIGRATION.md demands DB credentials (ANKER_JOB_DB_HOST/PORT/USER/PASSWORD/NAME/CHARSET) and a configs file. Requesting full DB credentials is proportionate to the task if the scripts are present, but it's not declared up front — this mismatch is risky. The skill also expects to surface raw 'request' payloads which may contain sensitive PII; that needs explicit handling and justification.
Persistence & Privilege
The skill does not request always:true and contains no install procedure that writes persistent agents-level configuration. run.sh simply execs Python scripts; there is no evidence it modifies other skills or system-wide settings.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install ak-data-daily-timeout-report - 安装完成后,直接呼叫该 Skill 的名称或使用
/ak-data-daily-timeout-report触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial public release for ClawHub.
元数据
常见问题
AK Data Daily Timeout Report 是什么?
Unified daily timeout entry for AK Data single job type: task timeout + crawler timeout + comparison + drill-down examples with log request. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 79 次。
如何安装 AK Data Daily Timeout Report?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ak-data-daily-timeout-report」即可一键安装,无需额外配置。
AK Data Daily Timeout Report 是免费的吗?
是的,AK Data Daily Timeout Report 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
AK Data Daily Timeout Report 支持哪些平台?
AK Data Daily Timeout Report 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 AK Data Daily Timeout Report?
由 LeeHoo29(@leehoo29)开发并维护,当前版本 v1.0.0。
推荐 Skills