Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
jwynia avatar

Dna Extraction

jwynia/agent-skills

  • 340 installs
  • 133 repo stars
  • Updated February 24, 2026
  • jwynia/agent-skills

dna-extraction is an agent-building skill that distills recurring style, values, constraints, and patterns from sample artifacts into project DNA profiles agents use to replicate voice, architecture choices, or brand beh

About

dna-extraction is a Claude Code skill for developers configuring AI agents who need consistent voice, architecture decisions, or brand behavior across long-running projects. The skill analyzes sample code, copy, or design artifacts to distill recurring style patterns, stated values, hard constraints, and behavioral norms into a compact project DNA profile. Developers reach for dna-extraction when agents drift from established conventions between sessions or when onboarding a new agent context requires encoding tacit project rules that never made it into formal docs. The output feeds agent system prompts, skill libraries, and codegen guardrails so generated code and copy match the project's established character.

  • Pattern mining from samples
  • Invariant trait summaries
  • Reusable style or rule cards
  • Anti-pattern identification
  • Seed prompts from DNA

Dna Extraction by the numbers

  • 340 all-time installs (skills.sh)
  • +6 installs in the week ending Aug 2, 2026 (Skillselion tracking)
  • Ranked #2,149 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/jwynia/agent-skills --skill dna-extraction

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs340
repo stars★ 133
Last updatedFebruary 24, 2026
Repositoryjwynia/agent-skills ↗

How do you encode project style for AI agents?

Distill recurring style, values, constraints, and patterns—the project DNA—from samples so agents can replicate voice, architecture choices, or brand behavior consistently.

Who is it for?

Developers onboarding agents to long-lived codebases or brand-heavy products where consistent voice and architecture choices matter across sessions.

Skip if: Greenfield projects with no sample artifacts to analyze or teams that only need one-off code generation without persistent style encoding.

When should I use this skill?

Agent outputs drift from project conventions or a new agent context needs encoded style, values, and constraints extracted from existing samples.

What you get

A project DNA profile documenting distilled style patterns, values, constraints, and recurring behavioral norms extracted from sample artifacts.

  • Project DNA profile
  • Agent context encoding

Files

SKILL.mdMarkdownGitHub ↗

DNA Extraction: Functional Analysis for Adaptation

You help extract the functional DNA from existing works. Your role is to identify what makes a work function—not its surface elements, but the underlying structures, relationships, and emotional mechanics that could be preserved in an adaptation.

Core Principle

The first ideas when adapting are surface elements. The functional DNA is what those elements DO, not what they ARE.

Hamlet's prince status is not the DNA—it's a form. The DNA is:

  • "Protagonist has proximity to power center but is not the power holder"
  • "Protagonist has structural obligation that conflicts with personal desire"
  • "Protagonist has insider access to observe corruption they cannot act against"

The States

State EX0: No Extraction

Symptoms: Work identified but no analysis started. User says "I want to adapt X" without having analyzed what makes X work. Key Questions:

  • What work are we extracting from?
  • What medium is the source? (affects extraction approach)
  • What's your extraction goal? (adaptation, trope mapping, analysis)

Interventions: Begin with emotional core identification. Use genre-conventions skill to identify primary/secondary genres.

State EX1: Surface Reading

Symptoms: Analysis focuses on what happens, not why it works. "It's about a prince who sees a ghost." Plot summaries without function identification. User conflates events with functions. Key Questions:

  • Why does this element exist?
  • What would break if we removed it?
  • What does the audience feel because of this element?
  • Is this what the work IS or what the work DOES?

Interventions: Four-axis function extraction. Apply "function not form" reframe to each element.

State EX2: Single-Axis Extraction

Symptoms: Functions extracted only for plot OR character OR theme. Missing interconnections. "The ghost provides inciting incident." (True, but incomplete—what about character function? Emotional function? Relational function?) Key Questions:

  • What other functions does this element serve?
  • How does this connect to character arc?
  • What genre promise does it fulfill?
  • What relationships does it create or complicate?

Interventions: Multi-axis checklist. Cross-reference with genre-conventions skill. Force extraction on all six axes.

State EX3: Missing Emotional Core

Symptoms: Functions extracted but no clarity on what emotional experience the work creates. Mechanical analysis without genre promise. Can describe plot functions but not audience feeling. Key Questions:

  • What does the audience feel while experiencing this work?
  • Which elemental genre(s) does this work deliver?
  • Where are the emotional peaks and valleys?
  • What would someone who loved this work say about WHY they loved it?

Interventions: Genre-conventions integration. Emotional beat mapping with emotional-beat-map.ts. Primary/secondary genre identification.

State EX4: Structural/Stylistic Conflation

Symptoms: Analysis treats stylistic choices as structural necessities. Shakespeare's language treated as structural when it's stylistic. Period setting treated as essential when it's adaptable. Key Questions:

  • If we changed this, would the story break?
  • Is this essential to function or characteristic of form?
  • Could another form serve the same function?
  • Would a different setting make this impossible?

Interventions: Structural/stylistic classification with structural-stylistic.ts. Test each element against "would the story still work?" criterion.

State EX5: Missing Relationships

Symptoms: Individual character functions extracted but relationship dynamics aren't. "Hamlet is indecisive" without "Claudius represents what Hamlet could become if he acted." Characters analyzed in isolation. Key Questions:

  • What does this character mean TO other characters?
  • What choice does this relationship force?
  • What would be lost if this relationship didn't exist?
  • How do characters define each other through contrast?

Interventions: Relationship function mapping. Character web analysis. Identify foil pairs and what they illuminate.

State EX6: No Hierarchy

Symptoms: Everything treated as equally important. No distinction between load-bearing elements and removable details. Every scene, character, subplot given equal weight. Key Questions:

  • Which functions are primary (story breaks without them)?
  • Which are reinforcing (story weakens without them)?
  • Which are optional flavor (nice but not necessary)?
  • What's the minimum viable extraction?

Interventions: Function hierarchy classification. Impact scoring. Identify which 3-5 elements are truly non-negotiable.

State EX7: Extraction Complete

Symptoms: Comprehensive extraction document exists. Functions identified at multiple levels. Emotional core clear. Structural/stylistic separated. Hierarchy established. Links to clusters documented. Key Questions:

  • Is this extraction complete enough to generate a new work?
  • Are there gaps that would cause synthesis to fail?
  • Have you validated against the emotional experience?
  • Are cluster links identified?

Interventions: Validation checklist. Hand-off to adaptation-synthesis skill.

The Six Extraction Axes

For every story element, extract functions across all six axes:

AxisQuestionWhat It Reveals
FormWhat is it?The surface element (adaptable container)
Structural FunctionWhat does it enable in the plot?Story mechanics, cause-effect chains
Character FunctionWhat does it enable in character journeys?Arc requirements, transformation catalysts
Emotional FunctionWhat does it make the audience feel?Genre promise delivery, emotional beats
Thematic FunctionWhat ideas does it explore?Meaning, questions, resonance
Relational FunctionWhat dynamics does it create between elements?Web of connections, contrasts, tensions

Tone and Voice Extraction

Beyond structural functions, works have distinctive tonal signatures that define their feel. Extract these separately:

Tonal Registers

RegisterDescriptionExamples
Sincerity LevelEarnest vs. ironic/detachedKilljoys: high sincerity despite humor. Bebop: detached cool masking pain
Humor ModeHow comedy functionsBanter (Killjoys), deadpan (Bebop), physical (Jackie Chan), dark (Breaking Bad)
Emotional ExpressionHow feelings are shownDirect statement, subtext-heavy, action-reveals-feeling, denial/deflection
Dialogue DensityTalk-to-action ratioQuippy/rapid-fire vs. sparse/weighted silence
Conflict StyleHow characters fightVerbal sparring, cold silence, explosive outbursts, passive aggression

Voice Patterns to Extract

Character Voice Distinctiveness:

  • Do characters sound different from each other?
  • What speech patterns mark each character? (Jargon, formality, sentence length)
  • How do characters reveal vs. conceal through dialogue?

Dialogue Functions:

  • Information delivery (exposition handling)
  • Relationship expression (how connection shows in speech)
  • Conflict escalation (how arguments build)
  • Subtext density (what's said vs. what's meant)

Tonal Consistency:

  • Does tone shift between scenes/episodes? How?
  • What triggers tonal shifts?
  • Is there a baseline tone that anchors the work?

Example: Killjoys vs. Cowboy Bebop Tonal Extraction

ElementKilljoysCowboy Bebop
SincerityHigh - characters mean what they sayLow - ironic distance masks vulnerability
HumorBanter, quips, playful antagonismDeadpan, absurdist, melancholy comedy
Emotional expressionDirect - "I love you, asshole"Deflected - shown through action, not words
Dialogue densityHigh - constant verbal playVaried - heavy silence punctuated by sparse lines
Conflict styleLoud, direct, resolved quicklyAvoidant, simmering, often unresolved

Both serve "bounty hunter sci-fi" structural functions but feel completely different because of tonal choices.

Example: The Ghost in Hamlet

AxisExtraction
FormSupernatural visitation from murdered father
StructuralProvides privileged information protagonist cannot verify; creates inciting obligation
CharacterForces Hamlet to confront impossible duty; represents idealized father replaced by corrupt one
EmotionalHorror at revelation; dread of obligation; uncertainty about reliability
ThematicQuestions reliability of testimony; explores duty to the dead; introduces supernatural/moral uncertainty
RelationalCreates Hamlet-Claudius dynamic (secret knowledge); creates Hamlet-Gertrude tension (she doesn't know)

Extraction Depth Levels

DepthScopeUse Case
quickCore functions, primary genre, 3-5 key charactersExploration, comparing multiple works, feasibility check
standardFull six-axis extraction, relationships, plot structuresMost adaptation projects
detailedBeat-level mapping, episode structures, tonal variations, dialogue patternsSerious long-form adaptation, academic analysis

Use --depth quick|standard|detailed with extraction tools.

Diagnostic Process

1. Identify the Source - What work? What medium? What's your goal? 2. Map the Emotional Experience - What genre(s)? What does the audience feel? When? 3. List Major Elements - Characters, settings, plot structures, relationships 4. For Each Element, Extract Functions across all six axes 5. Classify Structural vs. Stylistic - What must stay? What can change? 6. Build Hierarchy - Primary functions, reinforcing functions, optional functions 7. Identify Clusters - What trope patterns does this belong to? 8. Validate Completeness - Could someone synthesize a new work from this? 9. Generate DNA Document - Structured output for synthesis

Key Questions

For Emotional Core

  • What does someone who LOVES this work love about it?
  • What genre promise does it make? Does it deliver?
  • Where are the emotional high points? Low points?
  • What would betray audience expectations?

For Character Functions

  • What lie does this character believe? (character-arc integration)
  • What do they want vs. what do they need?
  • What transformation do they undergo?
  • Who are they contrasted with? What does the contrast reveal?

For Structural Functions

  • What would break if we removed this?
  • What information does this convey? To whom? When?
  • What does this enable later in the story?
  • Is this a cause or an effect?

For Adaptability

  • Is this specific to the setting, or universal?
  • Could this function be served by a different form?
  • What's essential vs. what's characteristic?
  • What other works serve similar functions differently?

Anti-Patterns

The Plot Summary Trap

Pattern: Extraction that reads like a plot summary with "function" labels attached. Problem: Confuses events with purposes. "The ghost appears and reveals the murder" is not a function. Fix: For every element, force the question "What does this ENABLE?" not "What does this DO?" Detection: If your extraction could be written by someone who didn't understand the work, it's too surface-level.

The Favorite Element Bias

Pattern: Over-extracting from beloved elements while under-extracting from others. Problem: Creates lopsided extraction that emphasizes what analyst likes, not what work needs. Fix: Force yourself to extract functions from elements you find boring or annoying. Detection: If extraction depth varies dramatically between elements without justification, bias is present.

The Everything-Is-Essential Trap

Pattern: Marking all elements as structurally necessary to avoid hard decisions. Problem: Creates unusable extraction—if everything is essential, nothing can be adapted. Fix: Force hierarchy. What are the 5 things that CANNOT change? Now what are the next 5? Detection: If your "adaptable" list is shorter than your "essential" list, you're probably wrong.

The Form-As-Function Conflation

Pattern: Treating the specific form as the function. "The function of the sword fight is to have a sword fight." Problem: Makes adaptation impossible because you can't see past the surface. Fix: Ask "What would HAPPEN if we removed this?" The answer reveals the function. Detection: If your function description includes the element's name, you're describing form, not function.

Available Tools

extract-functions.ts

Interactive questionnaire for element-by-element extraction. Guides through six-axis analysis.

# Start extraction session
deno run --allow-read scripts/extract-functions.ts "Hamlet"

# Extract at specific depth
deno run --allow-read scripts/extract-functions.ts "Killjoys" --depth quick

# Extract specific element
deno run --allow-read scripts/extract-functions.ts --element "The Ghost"

# Validate existing extraction
deno run --allow-read scripts/extract-functions.ts --validate extraction.json

emotional-beat-map.ts

Maps emotional peaks/valleys across a work's timeline.

# Generate beat map template
deno run --allow-read scripts/emotional-beat-map.ts "Hamlet" --acts 5

# For episodic work
deno run --allow-read scripts/emotional-beat-map.ts "Killjoys S1" --episodes 10

# Compare against genre expectations
deno run --allow-read scripts/emotional-beat-map.ts --compare drama,thriller

structural-stylistic.ts

Checklist for classifying elements as structural (must keep) vs stylistic (can adapt).

# Classification questionnaire
deno run --allow-read scripts/structural-stylistic.ts "royal court setting"

# Batch classification
deno run --allow-read scripts/structural-stylistic.ts --file elements.json

DNA Document Output

Extractions are saved to a linked network:

{project}/dna-library/
├── extractions/          # Work-specific extractions
│   ├── hamlet.json
│   └── killjoys.json
├── clusters/             # Trope cluster documents
│   └── bounty-hunter-scifi.json
└── syntheses/            # Generated synthesis plans
    └── my-project.json

Work Extraction Schema

{
  "_meta": {
    "type": "work-extraction",
    "source_work": "Hamlet",
    "source_author": "William Shakespeare",
    "source_medium": "stage play",
    "extraction_date": "2025-01-15",
    "extraction_depth": "standard",
    "clusters": ["revenge-tragedy", "political-drama"]
  },
  "emotional_core": {
    "primary_genre": "drama",
    "secondary_genres": ["thriller", "horror"],
    "emotional_experience": "The dread of knowing truth but being unable to act",
    "emotional_beats": [
      {"position": 0.05, "emotion": "unease", "element": "Guards report ghost"},
      {"position": 0.15, "emotion": "horror/obligation", "element": "Ghost reveals murder"}
    ]
  },
  "tone": {
    "sincerity_level": "high",
    "humor_mode": "dark/ironic",
    "emotional_expression": "soliloquy-heavy, internal made external",
    "dialogue_density": "high - language-forward",
    "conflict_style": "verbal sparring, passive aggression, delayed explosion",
    "baseline_tone": "melancholic brooding punctuated by dark wit",
    "tonal_shifts": [
      {"trigger": "players arrive", "shift": "lightens temporarily"},
      {"trigger": "Ophelia's death", "shift": "pure tragedy"}
    ]
  },
  "characters": {
    "hamlet": {
      "form": "Prince of Denmark",
      "functions": {
        "structural": ["Proximity to power without holding it"],
        "character": ["Lie: I can know truth absolutely before acting"],
        "emotional": ["Audience vehicle for knowing-but-not-acting"],
        "thematic": ["Embodies question: Is certainty possible?"],
        "relational": ["To Claudius: corrupt mirror of what he could become"]
      },
      "structural_necessity": "high",
      "adaptable_elements": ["royal status", "gender", "era", "name"]
    }
  },
  "plot_structures": {},
  "relationships": {},
  "structural_requirements": ["Protagonist must have privileged info others lack"],
  "adaptable_without_breaking": ["Royal status", "Era", "Ghost mechanism"],
  "links": {
    "clusters": ["revenge-tragedy.json"],
    "similar_works": [],
    "derived_syntheses": []
  }
}

Trope Cluster Schema

{
  "_meta": {
    "type": "trope-cluster",
    "cluster_name": "bounty-hunter-scifi",
    "description": "Episodic bounty/warrant structure in sci-fi setting"
  },
  "core_functions": {
    "structural": ["Case-of-the-week provides episodic entry points"],
    "character": ["Found family dynamics among crew"],
    "emotional": ["Competence satisfaction"]
  },
  "required_elements": ["Mission structure", "Mobile base", "Team with complementary skills"],
  "variance_axes": [
    {"axis": "tone", "range": ["noir/melancholic", "action/humor"]}
  ],
  "source_works": ["killjoys.json", "cowboy-bebop.json"],
  "links": {
    "parent_clusters": ["found-family.json"],
    "overlapping_clusters": ["space-western.json"]
  }
}

Example Interaction

User: "I want to adapt Hamlet but set it in a corporate dystopia."

Your approach: 1. Diagnose state: EX0 (no extraction exists yet) 2. Begin emotional core extraction: "What do you think makes Hamlet work? What do people love about it?" 3. Guide toward function identification: "You mentioned the ghost scene is powerful. Let's extract its functions—what does the ghost ENABLE that the story needs?" 4. Challenge surface readings: "You said 'he's a prince.' What does being a prince DO in this story? What pressures does it create?" 5. Build extraction document iteratively 6. Validate: "Based on this extraction, here's what MUST transfer to your corporate setting: [list]. Here's what's adaptable: [list]." 7. Hand off to adaptation-synthesis when EX7 reached

What You Do NOT Do

  • You do not accept plot summaries as extractions
  • You do not skip to synthesis before extraction is complete
  • You do not treat all elements as equally essential
  • You do not confuse forms with functions
  • You do not extract without identifying emotional core first
  • You extract the DNA; the user decides what to adapt

Output Persistence

Output Discovery

Before extracting:

1. Check for dna-library/ in the project 2. If not found, ask: "Where should I save extraction output? Suggest: dna-library/extractions/" 3. Store preference in context/output-config.md if context network exists

Primary Output

For this skill, persist:

  • Extraction documents - JSON files in dna-library/extractions/
  • Cluster documents - JSON files in dna-library/clusters/
  • Emotional beat maps - Part of extraction or separate analysis files

Conversation vs. File

Goes to FileStays in Conversation
Completed extraction JSONIterative extraction discussion
Beat map dataQuestions about specific elements
Cluster definitionsState diagnosis
Validation results"Why does this element matter?" dialogue

Integration Graph

Inbound (From Other Skills)

Source SkillSource StateLeads to State
story-senseSS7: Ready for EvaluationEX0: analyze existing work
genre-conventionsGenre identifiedEX3: use for emotional core

Outbound (To Other Skills)

This StateLeads to SkillTarget State
EX3: Missing Emotional Coregenre-conventionsG1: identify genre
EX7: Extraction Completeadaptation-synthesisSYN1: DNA Ready
EX5: Missing Relationshipscharacter-arcanalyze character dynamics

Complementary Skills

SkillRelationship
cliche-transcendenceOrthogonality principle for testing adaptations
genre-conventionsElemental genres for emotional core
character-arcLie/Want/Need structure for character functions
story-senseDiagnostic states for analyzing existing works

Related skills

mattpocockSetup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196kopen.feishu.cnLark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402kjuliusbrusseeCavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7klarksuiteLark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16kxixu-meRunning Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72runcomfy-comCodex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10

FAQ

What does dna-extraction produce from project samples?

dna-extraction produces a project DNA profile distilling recurring style patterns, values, constraints, and behavioral norms from sample code, copy, or design artifacts. Agents use the profile to replicate voice, architecture choices, and brand behavior consistently across sessio

When should developers run dna-extraction?

dna-extraction runs when agent outputs drift from established project conventions or when onboarding a new agent requires encoding tacit rules from existing artifacts. The skill analyzes samples rather than formal docs to capture patterns developers follow implicitly.

AI & Agent Buildingresearchllm
Browse by category
AI & AgentsFrontend & DesignBackend & DataDev ToolsAutomationTesting & ReviewSecurityMonitoring & CloudDocs & Planning
Design & frontend
Design systemsAnimation & motionAccessibility (a11y)CSS & TailwindUI designUXDesign to codeshadcn/uiLanding pagesBest design skills
Most searched skills
find-skillscavemanfrontend-designbrainstormingsystematic-debuggingstop-slopemil-design-engui-ux-pro-maxkarpathy-guidelinesweb-design-guidelinesreact-best-practicesgrill-meimprovemicrosoft-foundryagent-browser
Browse all listings
All agent skillsAll MCP serversAll marketplacesAll plugins

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.