PrivateInvestigator

Customization

Before executing, check for user customizations at: ~/​.claude/​skills/​CORE/​USER/​SKILLCUSTOMIZATIONS/​PrivateInvestigator/

If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.

PrivateInvestigator - Ethical People Finding

Core Principle

PUBLIC DATA ONLY - No hacking, pretexting, or authentication bypass. All techniques are legal and ethical.

Voice Notification

When executing a workflow, do BOTH:

1. Send voice notification:

   curl -s -X POST http://localhost:8888/​notify \
     -H "Content-Type: application/​json" \
     -d '{"message": "Running the WORKFLOWNAME workflow from the PrivateInvestigator skill"}' \
     > /​dev/​null 2>&1 &
   

2. Output text notification:

   Running the **WorkflowName** workflow from the **PrivateInvestigator** skill...
   

Full documentation: ~/​.claude/​skills/​CORE/​SkillNotifications.md

Workflow Routing

When executing a workflow, output this notification:

Running the **WorkflowName** workflow from the **PrivateInvestigator** skill...

When to Activate

Direct People-Finding

• "find [person]", "locate [person]", "search for [person]"
• "reconnect with [person]", "looking for lost contact"
• "find an old friend", "locate a former coworker"

Reverse Lookup

• "reverse phone lookup", "who owns this email"
• "reverse image search", "find person by username"

Investigation

• "background check" (public data only)
• "what can you find about [person]"
• "research [person]"

Available Workflows

Research Strategy

MANDATORY: Extensive Parallel Research

Every investigation uses 15 parallel research agents (5 types × 3 each):

Agent Types:

1. ClaudeResearcher (3 agents) - People search aggregators, professional records, location intelligence, comprehensive identity, public records, education/​alumni
2. GeminiResearcher (3 agents) - Alternative identities, multi-perspective synthesis, historical context
3. GrokResearcher (3 agents) - Social media deep search, contrarian analysis, real-time intelligence
4. CodexResearcher (3 agents) - Username enumeration, Google dorking, technical profiles

Each agent executes 3 sub-searches = 45 parallel search threads per investigation

Launch Pattern: All 15 agents launch in a SINGLE message with multiple Task tool calls.

Core Capabilities

1. People Search Aggregators

2. Social Media Investigation

• Facebook: Google x-ray searches, mutual friends, groups
• LinkedIn: Boolean search, alumni networks
• Instagram/​Twitter/​TikTok: Username patterns, cross-platform correlation

3. Public Records

• Voter Registration: Most states publicly available
• Property Records: County assessor/​recorder sites
• Court Records: PACER (federal), state court portals, CourtListener
• Business Filings: Secretary of State websites
• Professional Licenses: State licensing boards

4. Reverse Lookup

• Phone: CallerID, NumLookup, carrier lookup
• Email: Epieos, Holehe, Hunter.io
• Image: PimEyes, TinEye, Google/​Yandex Images
• Username: Sherlock, WhatsMyName, Namechk

5. Google Dorking

site:linkedin.com "John Smith" "Software Engineer"
site:facebook.com "lives in" "Austin" "marketing"
filetype:pdf resume "Jane Doe" "San Francisco"

Investigation Methodology

Information Hierarchy

Tier 1: Foundation Data

• Full name (and variations/​maiden names)
• Approximate age or date of birth
• Last known location
• Context (school, workplace, relationship)

Tier 2: Primary Research

• People search aggregators
• Social media presence scan
• Google dorking

Tier 3: Deep Investigation

• Public records searches
• Reverse lookups on discovered info
• Cross-platform correlation
• Associate/​family network mapping

Tier 4: Verification

• Multi-source confirmation
• Timeline consistency check
• Photo verification
• Confidence scoring

Confidence Scoring

Dealing with Common Names

1. Add Specificity - Include location, age, employer, school
2. Cross-Reference - Match DOB + address patterns across sources
3. Family Connections - Verify through known relatives
4. Timeline Analysis - Does the life history make sense?
5. Multiple Identifiers - Require 3+ matching data points

Legal & Ethical Boundaries

GREEN ZONE (Allowed)

✅ Search public records (property, court, voter, business) ✅ Access publicly posted social media content ✅ Use people search aggregator sites ✅ Perform reverse lookups on public data ✅ Google dorking with public search operators

RED ZONE (Never Cross)

❌ Access data behind login walls without authorization ❌ Bypass authentication or security measures ❌ Use pretexting or impersonation ❌ Access private databases (credit, financial, medical) ❌ Stalk, harass, or intimidate subjects ❌ Access PI-only databases without license

When to STOP

• If the purpose shifts to harassment or stalking
• If the subject has clearly opted out of contact
• If investigation requires illegal methods
• If you suspect the requestor has malicious intent

Examples

Example 1: Finding an Old College Friend

User: "Help me find my college roommate from 2005, John Smith from Austin"
→ Routes to FindPerson.md
→ Launches 15 parallel research agents
→ Cross-references people search + LinkedIn alumni + property records
→ Verifies identity through timeline analysis
→ Reports findings with HIGH confidence

Example 2: Reverse Phone Lookup

User: "Who called from 512-555-1234?"
→ Routes to ReverseLookup.md
→ Runs phone through CallerID, NumLookup
→ Cross-references with people search aggregators
→ Reports owner name, location, carrier

Example 3: Social Media Investigation

User: "Find Jane Doe's social media, she's a marketing professional in Denver"
→ Routes to SocialMediaSearch.md
→ LinkedIn Boolean search + Google x-ray
→ Username enumeration if handle discovered
→ Reports all accounts with MEDIUM/​HIGH confidence

---

Related Documentation:

• Complete workflow details in Workflows/ directory
• Integration with Research skill for parallel agent orchestration