返回提示词目录第 1374 页

数据库与存储桶暴露应急规划

梳理公开存储的封堵、证据保全和访问调查步骤,整理影响事实并提示通知义务需法务确认。

完整提示词

<context>
Exposed data stores are found constantly by automated scanners, so an open Elasticsearch, MongoDB or Redis instance, or a public bucket, should be assumed to have been seen. The questions that decide the outcome are how long it was open, whether anyone actually read or copied data, and what the data was. Those answers depend on logs that are often off by default: object-level read logging for buckets (S3 server access logs or CloudTrail data events, equivalent settings on other providers) and query logging for databases. Configuration change logs usually do exist and show when the exposure started. Unauthenticated databases are also wiped or ransomed by bots that leave a note collection or index; finding one proves access. Teams under pressure close the hole and delete the data, which destroys evidence and can turn an exposure into data loss. Personal data exposure may trigger legal notification duties with deadlines counted from awareness, which vary by jurisdiction and sector, so legal or privacy counsel should be involved early.
</context>

<task>
Plan the response:

<exposure>
{{exposure}}
</exposure>
{{#logs_available}}

<logs_available>
{{logs_available}}
</logs_available>
{{/logs_available}}
{{#data_description}}

<data_description>
{{data_description}}
</data_description>
{{/data_description}}

1. Close it: the quickest change that removes public access without deleting data (block public access settings, remove the public policy or ACL, bind the database to private addresses or add authentication, firewall rules), how to confirm from outside that it is closed, and who must approve. Record the time.
2. Preserve: export the current configuration, the change history, all available logs before they rotate, and a list of objects or indexes with sizes and timestamps. Note any ransom or wipe notes found.
3. How long was it open: use configuration change logs, infrastructure-as-code history and deployment records to find when the exposure began, and say how confident that is.
4. Who accessed it: analyse available access logs for reads from addresses outside your own, bulk listing or download patterns, and known scanners; if logs do not exist for the period, say plainly that access cannot be ruled out and what that means.
5. What data: classify what was in the store, count records and people affected, and identify especially sensitive fields (credentials, government ids, health, financial, children's data). Credentials or keys found must be rotated now.
6. Notification decision: summarise the facts counsel needs (data types, people, jurisdictions of those people, exposure window, evidence of access) and list the likely questions (regulator, individuals, customers under contract, the reporter). Mark any deadline you mention as a common example to verify, not a rule for this case.
7. Root cause and prevention: how the store became public (default setting, a temporary change never reverted, a misapplied template), and controls such as organisation-wide public access blocks, policy-as-code checks in CI, and scheduled external exposure scans. If an outside researcher reported it, reply and thank them.
</task>

<constraints>
- Never recommend deleting the exposed data or logs as part of containment.
- Do not decide whether notification is legally required; that is for legal or privacy counsel in the relevant jurisdictions.
- Do not state an absence of access when logs to prove it do not exist.
- Mark configuration changes as needing approval; never invent provider settings names.
{{> output/uncertainty}}
</constraints>

<output_format>
Use the seven sections as headings with numbered steps. Under "Who accessed it", give a table: Source | Activity | Evidence | Assessment. Under "Notification decision", give a fact sheet in a fenced block for counsel. End with a timeline table: Time | Event | Source.
</output_format>

可替换内容

  • {{exposure}}
  • {{#logs_available}}
  • {{logs_available}}
  • {{/logs_available}}
  • {{#data_description}}
  • {{data_description}}
  • {{/data_description}}
  • {{> output/uncertainty}}

分类与标签

编程开发、数据暴露、事件响应、日志取证、存储桶安全、泄露通知

前往 AI 图片创作