/extract-dna - Cognitive DNA Extraction

Extracts a person’s cognitive DNA across 5 knowledge layers: Philosophies, Mental Models, Heuristics, Frameworks, and Methodologies. Requires 3+ processed sources.

Syntax

/extract-dna [PERSON_NAME]

PERSON_NAME

string

required

Canonical person name (alex-hormozi, cole-gordon, sam-ovens, etc.)

Prerequisites

DNA extraction requires:

Dossier exists: knowledge/dossiers/persons/DOSSIER-{PERSON}.md
3+ sources processed: Minimum 3 files through /process-jarvis
Insights available: INSIGHTS-STATE.json has data for person
Chunks indexed: CHUNKS-STATE.json has chunks for person

The 5-Layer DNA Schema

Mega Brain structures expert knowledge into 5 distinct layers:

Layer 1: Philosophies
Layer 2: Mental Models
Layer 3: Heuristics
Layer 4: Frameworks
Layer 5: Methodologies

Core beliefs and worldview

Characteristics:
  - Fundamental beliefs
  - Appear 3+ times in different contexts
  - NO numeric thresholds
  - Shape decision-making framework

Example:
  - "Philosophy beats tactics"
  - "Speed matters more than perfection"
  - "Test everything, assume nothing"

Weight Calculation:
  Base: 0.50
  + 0.15 if direct quote with chunk_id
  + 0.15 if appears in 2+ sources
  + 0.10 if appears in 3+ sources
  + 0.05 if in narrative context
  - 0.20 if inferred

Thinking frameworks and lenses

Characteristics:
  - Change how you SEE, not what you DO
  - Generate specific questions
  - Analytical frameworks
  - Decision lenses

Example:
  - "Pareto Principle (80/20)"
  - "Inversion: What would cause failure?"
  - "Second-order thinking"

Weight Calculation:
  Same as Philosophies
  + 0.10 if prescriptive language

Practical decision shortcuts with thresholds

Characteristics:
  - MUST have numeric threshold
  - "If X then Y" format
  - Actionable rules
  - HIGHEST VALUE layer

Example:
  - "If show rate < 60%, email 12x in 48h"
  - "Close rate drops 40% without qualification"
  - "10% commission base + 5% accelerator at 80%"

Weight Calculation:
  Same as above
  + 0.10 if has specific threshold
  + 0.05 if prescriptive language

PRIORITY: Extract ALL heuristics with numbers

Named structures with components

Characteristics:
  - Named systems
  - Defined components
  - NO rigid order
  - Modular application

Example:
  - "STAR Qualification"
    - Situation
    - Timing
    - Authority
    - Resources
  
  - "3 Audience Buckets"
    - YES (ready buyers)
    - NO (not qualified)
    - MAYBE (nurture needed)

Weight Calculation:
  Same as above
  + 0.10 if all components defined

Step-by-step processes with rigid order

Characteristics:
  - RIGID sequence
  - Step-by-step
  - Success criteria per step
  - Can't skip or reorder

Example:
  - "Email Hammer (12 emails, 48h)"
    Step 1: Email 1-3 (Day 1 morning)
    Step 2: Email 4-6 (Day 1 afternoon)
    Step 3: Email 7-9 (Day 2 morning)
    Step 4: Email 10-12 (Day 2 afternoon)
    Success: >= 60% show rate

Weight Calculation:
  Same as above
  + 0.10 if success criteria defined

Execution Process

Step 1: Validate Prerequisites

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
DNA EXTRACTION: COLE GORDON
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

STEP 1: PRE-REQUISITE CHECK

[✓] DOSSIER exists: knowledge/dossiers/persons/DOSSIER-COLE-GORDON.md
[✓] INSIGHTS-STATE.json has insights for Cole Gordon
[✓] CHUNKS-STATE.json has chunks for Cole Gordon
[✓] Sources processed: 5 (CG001, CG002, CG003, CG004, CG005)

All prerequisites met. Proceeding...

If prerequisites missing:

[✗] EXTRACTION FAILED

Person "Sam Ovens" does not have sufficient data:
  ✗ DOSSIER: Not found
  ✗ Sources processed: 0 (minimum: 3)
  ✗ Insights: 0

Action required:
  1. Ingest 3+ materials for Sam Ovens
  2. Process each file with /process-jarvis
  3. Retry /extract-dna sam-ovens

Step 2: Load Source Data

STEP 2: LOADING SOURCES

• Insights: 247 found
• Narratives: 18 found
• Chunks: 487 indexed
• DOSSIER: 1,834 lines

Data loaded successfully.

Step 3: Extract by Layer

Extraction applies classification rules to identify items for each layer

STEP 3: EXTRACTING BY LAYER

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Layer 1: PHILOSOPHIES
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scanning for core beliefs (3+ occurrences, no numbers)...

  ✓ Extracted: 12 philosophies
  → Output: knowledge/dna/persons/cole-gordon/FILOSOFIAS.yaml

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Layer 2: MENTAL MODELS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scanning for analytical lenses...

  ✓ Extracted: 8 mental models
  → Output: knowledge/dna/persons/cole-gordon/MODELOS-MENTAIS.yaml

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Layer 3: HEURISTICS (PRIORITY LAYER)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scanning for rules with numeric thresholds...

  ✓ Extracted: 23 heuristics (18 with thresholds)
  → Output: knowledge/dna/persons/cole-gordon/HEURISTICAS.yaml

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Layer 4: FRAMEWORKS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scanning for named structures with components...

  ✓ Extracted: 15 frameworks
  → Output: knowledge/dna/persons/cole-gordon/FRAMEWORKS.yaml

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Layer 5: METHODOLOGIES
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scanning for step-by-step processes...

  ✓ Extracted: 7 methodologies
  → Output: knowledge/dna/persons/cole-gordon/METODOLOGIAS.yaml

Step 4: Calculate Weights

Each item gets a confidence weight:

Weight Formula:
  BASE: 0.50
  
  Bonuses:
    + 0.15  Direct quote with chunk_id
    + 0.15  Appears in 2+ sources
    + 0.10  Numeric threshold (heuristics only)
    + 0.10  Appears in 3+ sources
    + 0.05  Narrative context
    + 0.05  Prescriptive language
  
  Penalties:
    - 0.20  Inferred (not direct)
    - 0.15  Contradicts other item
    - 0.10  Ambiguous context

Usage Threshold: 0.70
  Items < 0.70 not used in agent responses

Example:

- id: "heuristic_cg_001"
  content: "If show rate drops below 60%, deploy Email Hammer (12 emails, 48h)"
  weight: 0.95
  calculation:
    base: 0.50
    + direct_quote: 0.15
    + two_sources: 0.15
    + numeric_threshold: 0.10
    + prescriptive: 0.05
    = 0.95
  sources: ["CG001_045", "CG003_089"]

Step 5: Generate CONFIG.yaml

# knowledge/dna/persons/cole-gordon/CONFIG.yaml

person:
  canonical_name: "Cole Gordon"
  company: "Closers.io"
  expertise: ["Sales", "Closing", "Team Building"]

sources:
  - CG001: "Masterclass - Scaling Sales Teams"
  - CG002: "Podcast - 28 Rules of Closing"
  - CG003: "Blueprint - Sales Process Design"
  - CG004: "VSL - High Ticket Closing"
  - CG005: "Course - Closers Academy Module 1"

dna_statistics:
  philosophies: 12
  mental_models: 8
  heuristics: 23
  frameworks: 15
  methodologies: 7
  total_items: 65
  avg_weight: 0.82
  items_above_threshold: 58 (89%)

behavioral_patterns:
  - "Emphasis on systems over talent"
  - "Data-driven decision making"
  - "Aggressive testing culture"
  - "Focus on show rate optimization"

extraction_metadata:
  extracted_at: "2026-03-06T10:45:00Z"
  insights_processed: 247
  chunks_referenced: 487
  version: "1.0.0"

Step 6: Completion Report

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
EXTRACTION COMPLETE: COLE GORDON
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

FILES GENERATED:
  • FILOSOFIAS.yaml      - 12 items
  • MODELOS-MENTAIS.yaml - 8 items
  • HEURISTICAS.yaml     - 23 items (18 with numeric thresholds)
  • FRAMEWORKS.yaml      - 15 items
  • METODOLOGIAS.yaml    - 7 items
  • CONFIG.yaml          - 1 file

STATISTICS:
  • Total items: 65
  • Items with weight >= 0.70: 58 (89%)
  • Insights processed: 247
  • Chunks referenced: 487

AVERAGE WEIGHT: 0.82

NEXT STEPS:
  1. Review HEURISTICAS.yaml (most valuable layer)
  2. Verify all items have chunk_ids
  3. Adjust weights if necessary
  4. View with: /view-dna cole-gordon

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

DNA File Structure

HEURISTICAS.yaml (Example)

# knowledge/dna/persons/cole-gordon/HEURISTICAS.yaml

heuristics:
  - id: "heuristic_cg_001"
    content: "If show rate drops below 60%, deploy Email Hammer (12 emails in 48h)"
    weight: 0.95
    threshold: "60%"
    action: "Email Hammer deployment"
    genealogy:
      insight_origin: "insight_CG003_042"
      chunks: ["chunk_CG003_089", "chunk_CG001_045"]
      sources: ["CG003", "CG001"]
    meta:
      category: "show-rate"
      actionable_by: ["LNS", "SALES-COORDINATOR"]
      priority: "HIGH"
  
  - id: "heuristic_cg_002"
    content: "Close rate drops 40% without proper qualification"
    weight: 0.88
    threshold: "40%"
    impact: "Close rate degradation"
    genealogy:
      insight_origin: "insight_CG002_023"
      chunks: ["chunk_CG002_067"]
      sources: ["CG002"]
    meta:
      category: "qualification"
      actionable_by: ["SDS", "CLOSER"]
      priority: "HIGH"
  
  # ... 21 more heuristics

metadata:
  person: "Cole Gordon"
  total_count: 23
  with_thresholds: 18
  avg_weight: 0.84
  extracted_at: "2026-03-06T10:45:00Z"

Examples

/extract-dna cole-gordon

Output Location

knowledge/dna/persons/{person-name}/
├── FILOSOFIAS.yaml        # Layer 1: Philosophies
├── MODELOS-MENTAIS.yaml   # Layer 2: Mental Models
├── HEURISTICAS.yaml       # Layer 3: Heuristics (MOST VALUABLE)
├── FRAMEWORKS.yaml        # Layer 4: Frameworks
├── METODOLOGIAS.yaml      # Layer 5: Methodologies
└── CONFIG.yaml            # Metadata and statistics

Genealogy Tracking

Every DNA item MUST trace back to source material

Genealogy Chain:
  DNA Item
  ↓
  insight_origin (insight ID)
  ↓
  chunks[] (chunk IDs)
  ↓
  sources[] (SOURCE_IDs)
  ↓
  Original file path in inbox/

Example:
  "Email Hammer" heuristic
  → insight_CG003_042
  → chunk_CG003_089
  → CG003
  → inbox/COLE GORDON/MASTERMINDS/scaling-masterclass.txt

Best Practices

1. Wait for 3+ Sources

Don’t extract DNA prematurely:

# Bad: Only 1 source processed
/extract-dna alex-hormozi
# Error: Insufficient data

# Good: 3+ sources processed
/ingest video1.mp4
/process-jarvis "inbox/ALEX HORMOZI/video1.txt"
/ingest video2.mp4
/process-jarvis "inbox/ALEX HORMOZI/video2.txt"
/ingest video3.mp4
/process-jarvis "inbox/ALEX HORMOZI/video3.txt"
/extract-dna alex-hormozi
# Success!

2. Prioritize Heuristics

Layer 3 (Heuristics) is the most valuable:

Why Heuristics Matter:
  - Actionable thresholds
  - Specific numbers
  - Immediate application
  - Highest ROI

After extraction, review:
  knowledge/dna/persons/{person}/HEURISTICAS.yaml

3. Verify Chunk IDs

All items must have chunk_ids:

# Check for missing chunk_ids
grep -r "chunks: \[\]" knowledge/dna/persons/cole-gordon/

# Should return nothing

4. Review Low-Weight Items

Items with weight < 0.70 need attention:

# Find low-weight items
yq '.heuristics[] | select(.weight < 0.70)' knowledge/dna/persons/cole-gordon/HEURISTICAS.yaml

# Either:
# - Find more evidence to boost weight
# - Remove if insufficient support

Troubleshooting

”Insufficient sources”

Issue: Less than 3 sources processed Solution:

# Check processed sources
cat system/SESSION-STATE.md | grep "Cole Gordon"

# Process more materials
/ingest video.mp4 --person "Cole Gordon"
/process-jarvis "inbox/COLE GORDON/video.txt"

“No heuristics found”

Issue: 0 heuristics extracted Solution: Heuristics require numeric thresholds. Check:

# Review insights for numbers
cat processing/insights/INSIGHTS-STATE.json | jq '.insights_state.persons["Cole Gordon"] | .[] | select(.content | test("[0-9]+"))'

# If none found, source materials may lack specific metrics
# Ingest more quantitative content (courses, frameworks)

“Missing chunk_ids”

Issue: DNA items without traceability Solution: This indicates a pipeline bug. Report it:

# Items without chunk_ids are invalid
# DO NOT use them in agent responses
# Re-run extraction after fixing pipeline

What Gets Created

Mind Clone Agent

DNA enables creation of agents/persons/{name}/ with SOUL + MEMORY

Agent Enrichment

Cargo agents get updated with person’s frameworks and heuristics

RAG Index

DNA files indexed for semantic search by Conclave

Dossier Updates

Person dossier annotated with DNA references

Next Steps

View DNA

Visualize extracted DNA schema

Conclave

Use DNA in multi-agent deliberations

DNA Concepts

Deep dive into the 5-layer model

Mind Clones

Create AI agents from DNA

Get Started

Core Concepts

CLI Commands

Guides

Advanced

​/extract-dna - Cognitive DNA Extraction

​Syntax

​Prerequisites

​The 5-Layer DNA Schema

​Execution Process

​Step 1: Validate Prerequisites

​Step 2: Load Source Data

​Step 3: Extract by Layer

​Step 4: Calculate Weights

​Step 5: Generate CONFIG.yaml

​Step 6: Completion Report

​DNA File Structure

​HEURISTICAS.yaml (Example)

​Examples

​Output Location

​Genealogy Tracking

​Best Practices

​1. Wait for 3+ Sources

​2. Prioritize Heuristics

​3. Verify Chunk IDs

​4. Review Low-Weight Items

​Troubleshooting

​”Insufficient sources”

​“No heuristics found”

​“Missing chunk_ids”

​What Gets Created

Mind Clone Agent

Agent Enrichment

RAG Index

Dossier Updates

​Next Steps

View DNA

Conclave

DNA Concepts

Mind Clones

Build docs developers (and LLMs) love

/extract-dna - Cognitive DNA Extraction

Syntax

Prerequisites

The 5-Layer DNA Schema

Execution Process

Step 1: Validate Prerequisites

Step 2: Load Source Data

Step 3: Extract by Layer

Step 4: Calculate Weights

Step 5: Generate CONFIG.yaml

Step 6: Completion Report

DNA File Structure

HEURISTICAS.yaml (Example)

Examples

Output Location

Genealogy Tracking

Best Practices

1. Wait for 3+ Sources

2. Prioritize Heuristics

3. Verify Chunk IDs

4. Review Low-Weight Items

Troubleshooting

”Insufficient sources”

“No heuristics found”

“Missing chunk_ids”

What Gets Created

Next Steps