服务等级协议 · Service Level Agreement · 05/10

服务等级协议Service Level Agreement

生效日期Effective 2026年2月1日
适用地区Jurisdiction 澳大利亚·全球Australia · Global
文件编号Doc No. MC-LEGAL-05
第一条Article 1

适用范围Scope

重要提示Important Notice

本 SLA 仅适用于 MiningClawd 付费订阅用户（专业版与企业版）。免费用户不享受 SLA 承诺。本协议中的"服务"特指云平台核心功能的实际可用性。
This SLA applies only to MiningClawd paid subscribers (Professional and Enterprise plans). Free users are not covered. 'Service' refers to the actual availability of core cloud platform functionalities.

1.1 定义 / 1.1 Definitions

服务 (Service)：指 MiningClawd 平台提供的所有核心功能，包括但不限于数据库查询、情报报告生成、3D 钻探可视化、AI 智能代理服务。
Service: Refers to all core functionalities provided by MiningClawd, including database queries, report generation, 3D drilling visualisation and AI Agent services.
宕机时间 (Downtime)：指服务因 MiningClawd 基础设施故障导致连续或间断无法访问的时间段。单次宕机事件必须持续 ≥1 分钟才计入统计。短暂网络波动（<1 分钟）不计为宕机。
Downtime: Refers to continuous or intermittent periods when Service is unavailable due to MiningClawd infrastructure failures. Individual incidents must last ≥1 minute to count. Transient network blips (<1 minute) are excluded.
计划维护 (Scheduled Maintenance)：指 MiningClawd 提前至少 72 小时通知的系统升级或维护工作。维护期间的服务中断不计为宕机。
Scheduled Maintenance: Refers to system upgrades or maintenance announced at least 72 hours in advance. Service interruptions during maintenance windows are excluded from Downtime calculation.

1.2 适用范围 / 1.2 Applicability

本 SLA 仅约束 MiningClawd 云平台的可用性承诺，包括：
This SLA covers only the availability commitments for the MiningClawd cloud platform, which includes:
以下项目不在 SLA 保障范围内：第三方集成（LinkedIn、WhatsApp、Email 网关）、手机应用原生功能的可用性、用户端网络连接质量。
The following are NOT covered: third-party integrations (LinkedIn, WhatsApp, Email gateways), mobile app native feature availability, user-side network quality.
第二条Article 2

服务可用性承诺Availability Commitment

2.1 正常运行时间目标 / 2.1 Uptime Targets

MiningClawd 承诺在每个自然月内提供以下正常运行时间保证：
MiningClawd commits to the following monthly uptime targets:
订阅计划
Plan
目标可用率
Target Uptime
允许最大宕机时间
Max Allowed Downtime
服务积分
Service Credit
专业版 (Professional)
Professional
 99.5%
约 3.6 小时/月
~3.6 hrs/month
<99%=5%月费/<99%=10% monthly fee; <95%=25% credit; <90%=50%
<99%=5% fee; <95%=25%; <90%=50%
企业版 (Enterprise)
Enterprise
 99.9%
约 43 分钟/月
~43 mins/month
<99.5%=10%月费/<99.5%=10% monthly fee; <99%=25%; <95%=50%
<99.5%=10%; <99%=25%; <95%=50%

2.2 正常运行时间计算公式 / 2.2 Uptime Calculation Formula

可用率计算公式: 可用率 = (当月总分钟数 − 停机分钟数) ÷ 当月总分钟数 × 100%。其中"当月总分钟数"为该自然月的全部分钟数（不扣除维护窗口），"停机分钟数"按第一条定义计算，即连续宕机 ≥1 分钟的时间累计。
Availability formula: Uptime = (Total minutes in month − Downtime minutes) ÷ Total minutes × 100%. "Total minutes" is the calendar month total (maintenance windows not deducted); "Downtime minutes" is cumulative duration of incidents lasting ≥1 minute as defined in Article 1.

2.3 可用性排除情形 / 2.3 Uptime Exclusions

下列情况导致的服务中断不计入正常运行时间目标的分母或分子计算：
Downtime caused by the following factors is excluded from uptime calculations:
第三条Article 3

计划维护Scheduled Maintenance

3.1 常规维护通知 / 3.1 Maintenance Notice

MiningClawd 将在计划维护前至少 72 小时通过以下渠道发送通知：
MiningClawd will provide notification at least 72 hours before planned maintenance via:

3.2 维护时间窗口 / 3.2 Maintenance Window

为最小化对全球用户的影响，计划维护按以下规格进行：
Planned maintenance is governed by the following specifications:
计划维护规格Scheduled Maintenance SLA

时间窗口：每月第二个周日 22:00-02:00 墨尔本时间（AEST/AEDT）
预通知：提前 14 天通过邮件通知
最长时长：单次维护窗口不超过 4 小时
年度累计：年度计划维护不超过 48 小时
排除：计划维护时间不计入可用率计算
Window: 2nd Sunday of each month, 22:00-02:00 Melbourne time (AEST/AEDT)
Notice: 14-day advance email notification
Duration: Single maintenance window max 4 hours
Annual limit: Cumulative annual scheduled maintenance shall not exceed 48 hours
Exclusion: Maintenance windows excluded from uptime calculation

在特殊情况下，维护时间可能调整至其他时段，但仍需遵守 14 天预通知原则。
In exceptional cases, maintenance may be rescheduled but must still comply with the 14-day advance notice requirement.

3.3 紧急维护 / 3.3 Emergency Maintenance

如发生影响平台安全或数据完整性的严重问题（如安全补丁），MiningClawd 有权立即进行紧急维护，可无需提前通知。在此情况下：
If a critical security or data integrity issue arises (e.g., security patches), MiningClawd may perform emergency maintenance immediately without advance notice. In such cases:
第四条Article 4

性能指标Performance Metrics

除可用性外，MiningClawd 还致力于维持以下性能指标。这些指标不直接触发服务积分，但作为服务质量的参考基准：
Beyond availability, MiningClawd maintains the following performance metrics. While not directly triggering service credits, these benchmark service quality:

4.1 API 响应时间 / 4.1 API Response Time

指标
Metric
目标
Target
说明
Notes
P50 延迟 < 200ms
中位数响应时间（全球平均）
Median response time (global average)
P95 延迟 < 1000ms
95th percentile 响应时间
95th percentile response time
P99 延迟 < 3000ms
99th percentile 响应时间
99th percentile response time

4.2 报告生成时间 / 4.2 Report Generation

标准情报报告生成时间：< 30 秒（从提交到用户接收的端到端时间，基于全球平均网络条件）
Standard intelligence report generation: < 30 seconds (end-to-end from submission to user receipt, global average conditions)

4.3 3D 可视化加载 / 4.3 3D Visualisation Loading

初始渲染时间：< 10 秒（从页面导航到首帧渲染，基于现代浏览器和正常网络连接）
Initial render: < 10 seconds (from navigation to first render, modern browser, normal connectivity)

4.4 AI 智能代理首字节延迟 / 4.4 AI Agent TTFB

首个文本令牌 (TTFB)：< 5 秒（从用户提交查询到 AI 代理返回首个响应令牌）
First token (TTFB): < 5 seconds (from user query submission to first response token from AI Agent)

4.5 AI 代理性能 SLA / 4.5 AI Agent Performance SLA

各 AI 代理的性能目标（最佳努力）如下：
Performance targets for each AI Agent (best-effort) are as follows:
代理名称
Agent
响应时间目标 (P99)
Latency Target (p99)
超时率上限
Timeout Limit
Lobby (市场分析)
Lobby (Market Analysis)
 < 10 秒 / < 10s < 0.5%
Crabby (监管合规)
Crabby (Regulatory)
 < 15 秒 / < 15s < 0.5%
Pincer (地质分析)
Pincer (Geology)
 3D 模型加载 < 5 秒 / < 5s model load < 1%
Clawdia (财务建模)
Clawdia (Financial)
 模型导出 < 30 秒 / < 30s export < 1%
重要提示Important Notice

AI 代理性能目标为最佳努力目标，不保证 AI 生成内容的准确性、真实性或适用性。详见 MC-LEGAL-09 AI 内容免责声明。
AI Agent performance targets are best-effort targets. No accuracy guarantee for AI-generated content applies. See MC-LEGAL-09 AI Content Disclaimer for details.

4.6 性能监控与报告 / 4.6 Performance Monitoring

MiningClawd 使用行业标准的实时监控工具（如 New Relic、DataDog）持续测量这些指标，并在月度报告中公开发布性能汇总数据。用户可随时在仪表板上访问近 12 个月的性能历史。
MiningClawd uses industry-standard real-time monitoring (e.g., New Relic, DataDog) to continuously measure these metrics and publishes performance summaries in monthly reports. Users can access 12-month performance history via their dashboard at any time.
第五条Article 5

事故响应Incident Response

5.1 严重等级分类与响应 SLA / 5.1 Severity Levels & Response SLA

MiningClawd 事故按下表分类，首次响应和修复目标时间如下：
MiningClawd classifies incidents with the following response and resolution targets:
严重程度
Severity
定义
Definition
首次响应
Initial Response
修复目标
Resolution Target
P1
关键
平台完全不可用
Platform completely unavailable
 15 分钟 / 15 min 4 小时 / 4 hours
P2
重大
核心功能受损（报告/代理不可用）
Core function impaired (reports/agents unavailable)
 30 分钟 / 30 min 8 小时 / 8 hours
P3
一般
非核心功能异常
Non-core function issue
 4 小时 / 4 hours 48 小时 / 48 hours
P4
低
界面/显示问题
UI/display issue
 1 工作日 / 1 business day 5 工作日 / 5 business days

5.2 首响应时间 (First Response SLA) / 5.2 First Response Time SLA

以下时间表定义了按计划和严重等级划分的首次响应时间承诺：
The following table defines first response time commitments by plan and severity:
严重等级
Severity
专业版
Professional
企业版
Enterprise
P1 2 小时 / 2 hrs 30 分钟 / 30 mins
P2 4 小时 / 4 hrs 2 小时 / 2 hrs
P3 8 小时 / 8 hrs 4 小时 / 4 hrs
P4 24 小时 / 24 hrs 12 小时 / 12 hrs

5.3 问题报告与升级 / 5.3 Issue Reporting & Escalation

用户可通过以下渠道报告问题：
Users can report issues via:
若初始响应未能令用户满意，用户可要求向技术主管进行升级。企业版用户可指定专属客户成功经理。
If the initial response is unsatisfactory, users may request escalation to a technical lead. Enterprise users may designate a dedicated Customer Success Manager.

5.4 沟通与更新 / 5.4 Communication During Incidents

在 P1 或 P2 级事故期间：
During P1 or P2 incidents:
第六条Article 6

服务积分Service Credits

6.1 积分计算与额度 / 6.1 Credit Calculation

若实际正常运行时间未达到承诺目标，MiningClawd 将根据下表向受影响的付费用户提供月度服务积分：
If actual uptime falls below the commitment, MiningClawd will issue monthly service credits per the table below to affected paid users:
月度正常运行时间
Monthly Uptime
服务积分
Service Credit
≥ 99.5%
无
None
99.0% 至 <99.5% 10%
95.0% 至 <99.0% 25%
< 95.0% 50%
积分计算公式：月度积分 = 月费 × (100 − 实际可用率%) ÷ 100。例如，月费为 USD 500 且实际可用率为 98.5%，则积分 = USD 500 × (100 − 98.5%) ÷ 100 = USD 7.50。
Credit calculation: Monthly Credit = Monthly Fee × (100 − Actual Uptime %) ÷ 100. For example, if monthly fee is USD 500 and actual uptime is 98.5%, credit = USD 500 × (100 − 98.5%) ÷ 100 = USD 7.50.

6.2 积分额度上限与申请流程 / 6.2 Credit Cap & Claim Process

服务积分为 SLA 违约的唯一补救措施。积分受以下限制和流程约束：
Service credits are the sole remedy for SLA breach. Credits are subject to the following limits and process:

6.3 积分的唯一救济 / 6.3 Exclusive Remedy

服务积分是用户因 SLA 未达成而享受的唯一救济措施。用户不得因 SLA 不达成而寻求其他赔偿、损害赔偿或退款，除非法律明确规定。积分放弃后不可恢复。
Service credits are the exclusive remedy for SLA shortfalls. Users may not seek additional compensation, damages, or refunds due to SLA failures unless expressly required by law. Forfeited credits cannot be recovered.
第七条Article 7

数据备份与恢复Data Backup & Recovery

7.1 备份策略 / 7.1 Backup Strategy

MiningClawd 采用以下数据保护和灾难恢复策略：
MiningClawd implements the following data protection and disaster recovery strategy:
备份与恢复指标Backup & Recovery Metrics

RPO (恢复点目标)：≤ 24 小时（最多丢失 24 小时内的数据）
RTO (恢复时间目标)：≤ 4 小时 (P1 级别)（从故障发生到系统恢复可用）
RPO (Recovery Point Objective): ≤ 24 hours (max 24-hour data loss)
RTO (Recovery Time Objective): ≤ 4 hours (P1 severity) (from incident to system recovery)

7.2 恢复目标 / 7.2 Recovery Objectives

指标
Metric
缩写
Abbreviation
目标
Target
说明
Definition
恢复点目标
Recovery Point Objective
 RPO ≤ 24 小时 / ≤ 24 hours
最多丢失数据量
Maximum data loss
恢复时间目标
Recovery Time Objective
 RTO ≤ 4 小时 / ≤ 4 hours
故障到恢复时间
Time to recover system
RPO (恢复点目标) ≤ 24 小时：意味着在发生灾难时，用户最多可能丢失 24 小时内的数据。
RTO (恢复时间目标) ≤ 4 小时：意味着从故障事件发生到系统恢复可用，目标为 4 小时以内。跨区域自动故障转移机制可加快恢复。
RPO ≤ 24 hours: In a disaster, users may lose up to 24 hours of data.
RTO ≤ 4 hours: Time from incident to full system recovery is targeted at 4 hours or less. Cross-region automatic failover accelerates recovery.

7.3 灾难恢复测试 / 7.3 Disaster Recovery Testing

MiningClawd 每年至少进行一次完整的灾难恢复演练（模拟），验证备份恢复流程的有效性。测试结果摘要将在年度安全审计报告中公布。
MiningClawd conducts at least one full disaster recovery drill annually to validate backup restore procedures. Test results summary will be published in the annual security audit report.

7.4 用户数据导出 / 7.4 User Data Export

用户可随时通过"数据导出"功能申请导出其帐户中的全部数据（数据库查询历史、下载的报告、AI 查询日志等）。导出请求应在 5 个工作日内完成，具体条款详见《数据导出条款》（MC-LEGAL-10）。
Users may request export of all account data (query history, downloaded reports, AI logs, etc.) via 'Data Export'. Requests are fulfilled within 5 business days per Data Export Terms (MC-LEGAL-10).
第八条Article 8

报告与透明度Reporting & Transparency

8.1 月度正常运行时间报告 / 8.1 Monthly Uptime Report

MiningClawd 在每个自然月结束后 10 个工作日内发布详细的月度报告，内容包括：
MiningClawd publishes a detailed monthly report within 10 business days of month-end, including:
月度报告将自动发送至所有活跃付费用户的注册邮箱，同时在用户账户仪表板上提供。
Monthly reports are automatically emailed to all active paid users and are accessible via their account dashboard.

8.2 季度业务评审（企业版） / 8.2 Quarterly Business Review (Enterprise)

MiningClawd 为所有企业版用户提供季度业务评审会议（QBR），由客户成功经理主持，内容包括：
MiningClawd provides quarterly business reviews (QBR) for all Enterprise users, led by a Customer Success Manager, covering:

8.3 公开状态页面 / 8.3 Public Status Page

MiningClawd 维护一个公开可访问的状态页面：status.miningclawd.com
MiningClawd maintains a publicly accessible status page: status.miningclawd.com
该页面实时显示：
This page displays in real time:
用户可订阅状态页面通知，在事件发生时收到自动警报。
Users can subscribe to status page alerts to receive automatic notifications of incidents.

8.4 事后报告 (Post-Incident Report) / 8.4 Post-Incident Report

对于所有持续时间 ≥ 30 分钟的宕机事件，MiningClawd 会在事件解决后 5 个工作日内发布详细的事后报告 (PIR)，包括：
For all incidents lasting ≥30 minutes, MiningClawd publishes a detailed post-incident report (PIR) within 5 business days of resolution, including:
PIR 将通过邮件发送至受影响用户，并在状态页面存档。
PIRs are emailed to affected users and archived on the status page.
第九条Article 9

排除情形Exclusions

9.1 不可抗力 / 9.1 Force Majeure

由下列原因引起的服务中断不计入 SLA 计算，且不触发服务积分：
Service interruptions caused by the following do not count toward SLA calculation and do not trigger credits:

9.2 第三方服务中断 / 9.2 Third-Party Outages

由 MiningClawd 所依赖的第三方供应商（如 AWS、Cloudflare、DNS 服务商）导致的中断通常不计入 SLA 目标，但 MiningClawd 有义务：
Outages caused by third-party vendors that MiningClawd depends on (AWS, Cloudflare, DNS providers) typically do not count toward SLA targets, but MiningClawd must:

9.3 用户侧故障或滥用 / 9.3 User-Side Issues & Abuse

下列情况不计入 SLA：
The following do not count toward SLA:

9.4 测试版功能与预览功能 / 9.4 Beta & Preview Features

任何明确标记为"Beta"、"Preview"或"Experimental"的功能不在 SLA 保障范围内。这些功能可能随时中断、改变或删除，不触发服务积分。
Any feature explicitly marked as 'Beta', 'Preview' or 'Experimental' is NOT covered by this SLA. Such features may be interrupted, changed, or removed at any time without triggering credits.

9.5 已弃用功能 / 9.5 Deprecated Features

MiningClawd 已公开宣布弃用的功能在弃用期间不适用 SLA。弃用的功能将在计划停用日期前至少 12 周通知用户。
Features publicly announced as deprecated do not have SLA coverage during the deprecation period. Deprecation is announced at least 12 weeks before planned discontinuation.
第十条Article 10

联系我们Contact Us

关于 SLA 和服务质量的任何问题或投诉，请通过以下渠道与我们联系：
For questions or complaints regarding this SLA and service quality, please contact us via:
事项
Matter
联系方式
Contact
响应时间
Response Time
SLA 相关咨询
SLA-related enquiries
 info@lynaimining.com
1 工作日
1 business day
一般支持和故障报告
General support & incident reports
 info@lynaimining.com
取决于严重等级
Depends on severity
P1 级紧急情况（24/7）
P1 emergencies (24/7)
 info@lynaimining.com
30 分钟内
Within 30 mins
系统状态查询
System status checks
 status.miningclawd.com
实时
Real-time
邮寄地址（法律通知）
Postal address (legal notices)
Level 10, 123 Collins Street, Melbourne VIC 3000, Australia
Level 10, 123 Collins Street, Melbourne VIC 3000, Australia
 —
免责声明Disclaimer

本 SLA 中的承诺基于目前的技术能力。MiningClawd 保留在不违反现有用户协议前提下更新或修改 SLA 条款的权利。任何重大修改将提前 30 天通知。
Commitments in this SLA reflect our current technical capabilities. MiningClawd reserves the right to update or modify SLA terms without breaching existing user agreements. Significant changes will be communicated 30 days in advance.