‹ 首页

triage-suspicious-login

@dandye · 收录于 5 天前 · 上游提交 3 个月前

Triage suspicious login alerts like impossible travel, untrusted location, or multiple failures. Use when investigating authentication anomalies. Analyzes user history, source IP reputation, login patterns, and determines if escalation is needed.

适合你,如果经常需要处理异常登录告警并快速决策

/ 通过 npx 安装 校验哈希
npx oh-my-skill add dandye/ai-runbooks/triage-suspicious-login
/ 通过 bash 安装
curl -fsSL https://oh-my-skill.com/install.sh | bash -s -- dandye/ai-runbooks/triage-suspicious-login
/ 已经装过?验证本机副本,不用重装
npx oh-my-skill verify dandye/ai-runbooks/triage-suspicious-login
安装目标可用 --agent / --scope 或 --to 明确指定;省略时只会在唯一已存在的 agent 目录上自动选择,零命中或多命中会停止并提示。content_hash 缺失或不一致均拒装。
119GitHub stars
~783上下文体积 · 单文件
索引托管

怎么用

商店整理自技能原文 · 版本 086cbf6 · 表述以原文为准
它做什么

装上后,Claude 会主动调查可疑登录告警:提取用户、源IP等信息,分析登录模式,判断是合法、可疑还是恶意,给出风险分数和升级建议。

什么时候触发

当你提供可疑登录案件的 ID 或相关标识(如用户ID、源IP)时触发。

装好后可以这样说
Claude 将开始分流流程。
跳过案件查找,直接分析。
聚焦近期登录活动分析。
技能原文 SKILL.md作者撰写 · Apache-2.0 · 086cbf6

Suspicious Login Triage Skill

Guide initial triage of suspicious login alerts (impossible travel, untrusted location, multiple failed logins) for Tier 1 SOC Analysts.

Inputs
  • CASE_ID - SOAR case ID containing the alert(s)
  • ALERT_GROUP_IDENTIFIERS - Alert group identifiers from the case
  • (Optional) USER_ID - The user ID if known upfront
  • (Optional) SOURCE_IP - The source IP if known upfront
Workflow
Step 1: Get Case Context
secops-soar.get_case_full_details(case_id=CASE_ID)
Step 2: Extract Key Entities
secops-soar.list_events_by_alert(case_id=CASE_ID, alert_id=ALERT_ID)

Parse events to extract:

  • USER_ID - The user account
  • SOURCE_IP - The login source IP
  • HOSTNAME - The target/source hostname (if available)
Step 3: User Context (SIEM)
secops-mcp.lookup_entity(entity_value=USER_ID)

Record: Recent activity, first/last seen, related alerts.

Step 4: Source IP Enrichment

Use /enrich-ioc with IOC_TYPE="IP Address":

  • GTI reputation and geolocation
  • SIEM entity summary
  • IOC match status
Step 5: Hostname Context (if available)
secops-mcp.lookup_entity(entity_value=HOSTNAME)
Step 6: Recent Login Activity

Search for login patterns over the last 96 hours:

secops-mcp.search_security_events(
    text='metadata.event_type IN ("USER_LOGIN", "AUTH_ATTEMPT") AND principal.user.userid = "USER_ID"',
    hours_back=96
)

Analyze for:

  • Logins from unusual IPs
  • Successful logins after failures
  • Geographic anomalies (impossible travel)
  • Concurrent sessions from different locations
Step 7: Check Related Cases

Use /find-relevant-case with search terms: [USER_ID, SOURCE_IP, HOSTNAME]

Step 8: (Optional) Identity Provider Check

If IDP tools available (e.g., Okta):

  • Account status
  • MFA enrollment
  • Recent legitimate logins
  • Password change history
Step 9: Synthesize & Document

Use /document-in-case with findings summary:

Suspicious Login Triage for USER_ID from SOURCE_IP:
- User SIEM Summary: [...]
- Source IP GTI: [reputation, geo]
- Login Pattern: [normal/anomalous]
- Related Cases: [...]
- Recommendation: [Close as FP | Escalate to Tier 2]
Required Outputs

After completing this skill, you MUST report these outputs:

| Output | Description | |--------|-------------| | LOGIN_VERDICT | Assessment: legitimate, suspicious, or malicious | | ANOMALY_INDICATORS | What made the login suspicious (impossible travel, new device, etc.) | | RELATED_ACTIVITY | Other suspicious activity from user or source IP | | RISK_SCORE | Numerical risk assessment (0-100) based on findings |

Decision Matrix

| Finding | Recommendation | |---------|----------------| | Known VPN/corporate IP + normal pattern | Close as FP | | User confirmed travel + MFA used | Close as Benign TP | | Malicious IP reputation | Escalate | | Impossible travel + no MFA | Escalate urgently | | Multiple failures then success from new IP | Escalate | | Pattern matches user's normal behavior | Close as FP |

Key Patterns to Detect

Impossible Travel:

  • Login from NYC, then London 30 mins later
  • Check if VPN or cloud service could explain

Credential Stuffing:

  • Many failures across multiple accounts from same IP
  • Success after many failures

Account Takeover:

  • Login from new device/location
  • Followed by password change or MFA modification

Lateral Movement:

  • Same user logging into many systems rapidly
  • Unusual service account activity
按 Apache-2.0 许可原样转载,未经改动 · 在 GitHub 查看 →

评论

登录即可评论;带「已验证安装」的,是发布者名下有本店的安装或持有记录。