external/anthropic-cybersecurity-skills/skills/performing-insider-threat-investigation/SKILL.md
Investigates insider threat incidents involving employees, contractors, or trusted partners who misuse authorized access to steal data, sabotage systems, or violate security policies. Combines digital forensics, user behavior analytics, and HR/legal coordination to build an evidence-based case. Activates for requests involving insider threat investigation, employee data theft, privilege misuse, user behavior anomaly, or internal threat detection.
npx skillsauth add seikaikyo/dash-skills performing-insider-threat-investigationInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Do not use for external attacker investigations where compromised credentials are used without insider collusion; use standard incident response procedures instead.
Document the initial report and validate before proceeding:
Investigation Authorization:
━━━━━━━━━━━━━━━━━━━━━━━━━━━
Case ID: INV-2025-042
Subject: [Employee Name] - [Title] - [Department]
Allegation: Unauthorized transfer of proprietary data to personal cloud storage
Reported By: DLP system alert + manager concern
Legal Approval: [Counsel Name] - 2025-11-15
HR Liaison: [HR Name]
Scope: File access and transfer activity from 2025-10-01 to present
Systems in Scope: Workstation, email, cloud storage, VPN, DLP logs
Gather evidence without alerting the subject to the investigation:
Log-Based Evidence (non-intrusive):
User Activity Monitoring (requires legal approval):
Endpoint Forensics (if warranted by evidence):
Build a behavioral profile comparing normal vs. anomalous activity:
Behavioral Analysis:
━━━━━━━━━━━━━━━━━━
Normal Baseline (6-month average):
- Login time: 08:30-09:00 weekdays
- Files accessed: 15-25 per day (marketing department files)
- Email volume: 45 sent, 80 received per day
- Data transferred: 50MB per day average
- USB usage: None
Investigation Period (last 30 days):
- Login time: 22:00-02:00 (after hours, multiple occasions)
- Files accessed: 200+ per day (finance, engineering, executive files)
- Email volume: 120 sent per day (30% to personal gmail)
- Data transferred: 2.5GB per day average
- USB usage: 3 unique devices connected (Kingston DataTraveler)
- Print jobs: 847 pages (competitor analysis, customer lists, source code)
Anomaly Score: 94/100 (Critical)
Build a chronological timeline of the subject's actions:
Timeline of Activity:
2025-10-15 Subject submits resignation (2-week notice)
2025-10-16 First after-hours login at 23:15, accessed engineering Git repository
2025-10-17 USB device (Kingston DT 64GB) first connected at 23:30
2025-10-18 DLP alert: 450 files copied to USB, including CAD drawings
2025-10-19 200+ emails forwarded to personal Gmail account
2025-10-20 Google Drive desktop client installed, syncing corporate SharePoint
2025-10-22 Accessed executive SharePoint site (not normally accessed)
2025-10-25 Second USB device connected, 2.1GB transferred
2025-10-28 Print job: 847 pages including customer contact database
Evaluate the severity and coordinate the response with HR and legal:
Impact Assessment:
Response Options (determined by legal and HR):
Ensure all evidence meets legal admissibility standards:
| Term | Definition | |------|------------| | Insider Threat | Risk posed by individuals with authorized access who intentionally or unintentionally cause harm to the organization | | User Behavior Analytics (UBA) | Technology that analyzes user activity patterns to detect anomalies indicating potential insider threats | | Data Loss Prevention (DLP) | Technology that monitors, detects, and blocks unauthorized transfer of sensitive data outside the organization | | Legal Hold | Directive to preserve all relevant evidence and suspend normal document destruction policies during an investigation | | Need to Know | Information access principle restricting insider threat investigation details to only authorized team members | | Exfiltration Vector | Method used to move data outside the organization: USB, email, cloud storage, print, screen capture, photography |
Context: A senior software engineer with access to critical repositories submits a two-week resignation notice. The engineering manager reports that the engineer has been working unusual hours and downloading large amounts of code.
Approach:
Pitfalls:
INSIDER THREAT INVESTIGATION REPORT
=====================================
Case ID: INV-2025-042
Classification: CONFIDENTIAL - Need to Know Only
Subject: [Name Redacted] - Senior Engineer
Investigation Period: 2025-10-01 to 2025-10-28
Investigator: [Name]
Legal Counsel: [Name]
HR Liaison: [Name]
ALLEGATION
Unauthorized exfiltration of proprietary source code and customer
data following resignation submission.
EVIDENCE SUMMARY
1. Git logs: 47 repositories cloned (vs. baseline of 3)
2. USB transfers: 4.6 GB across 3 unique devices over 12 sessions
3. Email: 200+ emails with attachments forwarded to personal Gmail
4. Cloud: Google Drive sync client installed, syncing corporate files
5. Print: 847 pages including customer contact database
6. Physical access: After-hours badge access on 8 of 12 workdays
BEHAVIORAL ANALYSIS
[Baseline vs. anomalous activity comparison]
IMPACT ASSESSMENT
Data Classification: Confidential (source code, customer PII)
Estimated Volume: 7.2 GB exfiltrated
Regulatory Impact: Potential GDPR notification (customer PII)
Business Impact: Competitive advantage at risk
TIMELINE
[Chronological event listing]
RECOMMENDATIONS
1. [Legal/HR decision on employment action]
2. [Evidence preservation actions]
3. [Regulatory notification assessment]
4. [Access control improvements]
development
拋棄式 HTML mockup 比稿:產出 2 到 3 個設計立場不同的變體(密度 / 版式 / 強調軸,不是換色),各附取捨說明,最後給有立場的對比結論。適用:「畫個草圖」「比較 A 版 B 版」「先看方向再做」「給我看幾種做法」。要 production 元件或設計已定案時不適用。
tools
需求不明時的意圖萃取訪談:一次一題、每題附上自己的猜測、聽出「真正想要 vs 覺得應該要」,直到能預測使用者反應(約 95% 信心)才動工。適用:需求缺少對象 / 動機 / 成功標準 / 約束,或使用者點名「訪談我」「先確認一下」「我們確定嗎」。明確自足的指示、純資訊查詢、機械性操作不適用。
development
對非平凡決策啟動新鮮 context 對抗審查(找碴不背書),在修正還便宜的時候抓出錯誤方向。適用:高風險改動(production、資安敏感邏輯、不可逆操作)、不熟的程式碼、要宣稱「這樣是安全的 / 可行的」之前。機械性操作與一行修改不適用。
testing
Reference for writing and editing agent skills well — the vocabulary and principles that make a skill predictable. Consult when authoring, reviewing, or pruning a SKILL.md.