Intel Workbench

Structured analytic techniques for cyber threat intelligence.

Intel Workbench is an interactive Analysis of Competing Hypotheses (ACH) tool that brings rigorous intelligence methodology to the browser. Score evidence against hypotheses, map findings to MITRE ATT&CK, identify cognitive biases, and export structured assessments. Zero backend, full offline capability, and eight distinct visual themes.

---

✨ Features

• ACH Matrix : Interactive evidence-vs-hypothesis grid with consistency ratings (C/I/N/NA), weighted scoring, and automatic preferred-hypothesis identification
• MITRE ATT&CK Tagging : Tag evidence and hypotheses with techniques from the Enterprise ATT&CK matrix (691 techniques, 14 tactics). Searchable by ID, name, or tactic. Vendored locally so the workbench stays offline-first
• Cognitive Bias Checklist : Heuer & Pherson taxonomy with 12 biases across Cognitive, Analytical, and Social categories; track mitigation notes per bias
• Score Visualization : Real-time normalized score bars showing hypothesis support levels with color-coded confidence indicators
• ICD 203 Estimative Language : Pick a likelihood band ("almost no chance" through "almost certainly") with the canonical 1-5%/5-20%/.../95-99% ranges per ODNI Analytic Standards; the preferred hypothesis displays a probability ribbon on the matrix and in Markdown exports
• Evidence Weighting : Credibility and relevance ratings (High/Medium/Low) that feed into weighted inconsistency scores
• Export & Import : Full JSON export/import for backup and sharing; Markdown export for report generation (includes ATT&CK technique IDs)
• 8 Visual Themes : Langley, Terminal, Analyst's Desk, Stratcom, Cyber Noir, Casefile Atlas, Ops Floor, and Blacksite Minimal
• In-App Guided Tour : First-visit walkthrough powered by driver.js highlighting every major feature
• Built-In Documentation : Comprehensive help page covering ACH methodology, scoring, bias awareness, and keyboard shortcuts
• Offline-First : All data persisted in localStorage; works without any server
• Keyboard Accessible : Full keyboard navigation across the matrix grid

---

🏗️ Architecture

Intel Workbench is a single-page React application with no backend dependencies:

Browser
  └─ React 18 (SPA, React Router v6)
       ├─ Zustand Store ← persist middleware → localStorage
       ├─ ThemeContext (per-variant color tokens)
       ├─ Pages: Home / ACH / Bias / Export / Docs
       └─ 8 Variant Layouts (lazy-loaded)

• State Management: Zustand with persist middleware writes to localStorage under the key intel-workbench-projects
• Routing: React Router v6 with nested variant routes (/v1/*, /v2/*, …, /v8/*, /default/*) and a variant picker at /
• Theming: ThemeContext provides color tokens per variant; components read them via useTheme()
• Code Splitting: Variant layouts are React.lazy() loaded to keep the initial bundle small

---

🚀 Quick Start

Prerequisites

• Node.js ≥ 18
• npm ≥ 9

Install & Run

git clone https://github.com/solomonneas/intel-workbench.git
cd intel-workbench
npm install
npm run dev

Open http://localhost:5173 in your browser.

Build for Production

npm run build
npm run preview

Run Tests

npm test         # vitest, single run
npm run test:watch
npm run typecheck

CI runs typecheck + tests + production build on every push (.github/workflows/ci.yml).

---

🛠️ Tech Stack

Layer	Technology	Purpose
Framework	React 18	Component UI
Language	TypeScript 5	Type safety
Styling	Tailwind CSS 3	Utility-first CSS
State	Zustand 4	Global state + persistence
Routing	React Router 6	Client-side navigation
Icons	Lucide React	Consistent icon set
Bundler	Vite 7	Dev server + build
Tour	driver.js 1.3 (CDN)	Guided onboarding

---

📁 Project Structure

intel-workbench/
├── index.html                 # Entry point + CDN links
├── package.json
├── vite.config.ts
├── tailwind.config.js
├── tsconfig.json
├── public/
│   └── vite.svg
└── src/
    ├── main.tsx               # React root
    ├── App.tsx                # Router + variant routes
    ├── index.css              # Tailwind layers + component classes
    ├── components/
    │   ├── ach/
    │   │   ├── ACHMatrix.tsx  # Interactive hypothesis matrix
    │   │   └── ACHScoreBar.tsx
    │   ├── bias/
    │   │   └── BiasChecklist.tsx
    │   ├── layout/
    │   │   └── AppShell.tsx   # Default sidebar layout
    │   └── GuidedTour.tsx     # driver.js onboarding tour
    ├── contexts/
    │   └── ThemeContext.tsx    # Theme color provider
    ├── data/
    │   ├── biasData.ts        # Cognitive bias catalog
    │   └── sampleProject.ts   # Sandworm sample data
    ├── pages/
    │   ├── HomePage.tsx       # Project list & creation
    │   ├── ACHPage.tsx        # Matrix workspace
    │   ├── BiasPage.tsx       # Bias review
    │   ├── ExportPage.tsx     # JSON/Markdown export
    │   ├── DocsPage.tsx       # In-app help & documentation
    │   └── VariantPicker.tsx  # Theme selector landing
    ├── store/
    │   └── useProjectStore.ts # Zustand store (persisted)
    ├── types/
    │   └── index.ts           # TypeScript interfaces
    ├── utils/
    │   ├── achScoring.ts      # Scoring algorithms
    │   ├── id.ts              # ID generator
    │   └── useBasePath.ts     # Variant-aware navigation
    └── variants/
        ├── v1/Layout.tsx      # Langley (intel agency)
        ├── v2/Layout.tsx      # Terminal (hacker)
        ├── v3/Layout.tsx      # Analyst's Desk (clean)
        ├── v4/Layout.tsx      # Stratcom (military)
        ├── v5/Layout.tsx      # Cyber Noir (cyberpunk)
        ├── v6/Layout.tsx      # Casefile Atlas (evidence desk)
        ├── v7/Layout.tsx      # Ops Floor (live cell)
        └── v8/Layout.tsx      # Blacksite Minimal (brutalist)

---

🎨 8 Variants

Each variant wraps the same core pages in a unique visual identity:

Variant	Theme	Aesthetic
v1 : Langley	Intelligence Agency	Dark navy, gold accents, serif type, classified stamps
v2 : Terminal	Hacker / OSINT	Pure black, matrix green, scanline overlay, monospace
v3 : Analyst's Desk	Clean Professional	Light backgrounds, blue accents, content-first layout
v4 : Stratcom	Military Command	OD green, amber accents, grid patterns, military time
v5 : Cyber Noir	Cyberpunk	Neon cyan + magenta, glow effects, glass-morphism
v6 : Casefile Atlas	Evidence Desk	Warm paper, red-thread evidence board, serif-heavy dossiers
v7 : Ops Floor	Live Cell	Dense command-center layout, teal signal lines, amber status blocks
v8 : Blacksite Minimal	Brutalist	Severe monochrome, acid-lime emphasis, hard-edged controls

All variants share the same Zustand store and page components. Switching themes is instant : just navigate back to the variant picker at /.

---

🎯 MITRE ATT&CK Integration

Tag evidence and hypotheses with techniques from the MITRE ATT&CK Enterprise matrix. Search by technique ID (T1059), name (Phishing), or filter by tactic (Initial Access, Execution, Lateral Movement, …). Tags persist in JSON exports and are rendered as clickable references in Markdown reports.

The full ATT&CK Enterprise dataset (691 techniques, 14 tactics) is vendored at src/data/attack-enterprise.json and lazy-loaded so the initial bundle stays small. To refresh after a new ATT&CK release:

curl -sL https://raw.githubusercontent.com/mitre/cti/master/enterprise-attack/enterprise-attack.json \
  | jq -f scripts/slim-attack.jq > src/data/attack-enterprise.json

---

📊 ACH Methodology

Analysis of Competing Hypotheses (ACH) is a structured analytic technique developed by Richards J. Heuer Jr. at the CIA. Instead of seeking evidence to confirm a preferred hypothesis, ACH forces analysts to:

1. Enumerate all reasonable hypotheses
2. List all significant evidence and arguments
3. Rate each evidence item against each hypothesis as Consistent (C), Inconsistent (I), Neutral (N), or Not Applicable (NA)
4. Score inconsistencies : the hypothesis with the fewest weighted inconsistencies is the most supported
5. Identify and mitigate cognitive biases that might distort the analysis

The key insight: disprove rather than prove. A single strong inconsistency can eliminate a hypothesis, while consistent evidence alone cannot confirm one.

Scoring Formula

Score = Σ (weight × rating_value)

where:
  rating_value: I = +2, N = 0, C = −1
  weight:       credibility_multiplier × relevance_multiplier
  multipliers:  High = 1.5, Medium = 1.0, Low = 0.5

Lower (more negative) scores indicate stronger support. The hypothesis with the lowest score is flagged as preferred.

---

📄 License

MIT : see LICENSE for details.