ThreatPrism

Multi-framework threat intelligence for AI coding agents

ThreatPrism is an MCP (Model Context Protocol) server that provides simultaneous threat analysis across four security frameworks — STRIDE, DREAD, LINDDUN, and PASTA — with automatic cross-referencing to CWE and MITRE ATT&CK.

Unlike single-framework tools, ThreatPrism gives you a multi-dimensional view of every threat: what category (STRIDE), how severe (DREAD), what privacy impact (LINDDUN), and what attack process (PASTA), all correlated in one analysis.

---

What Makes This Different

Capability	ThreatPrism	Typical Security Tools
Multi-framework correlation	STRIDE + DREAD + LINDDUN + PASTA in one pass	Usually one framework
Quantitative scoring	DREAD 1-10 scores with weighted context	Qualitative High/Med/Low
Privacy-first analysis	Built-in LINDDUN engine	Usually separate DPIA tool
Attack tree generation	AND/OR decomposition with likelihood	Manual diagramming
CWE + MITRE ATT&CK mapping	Automatic cross-reference	Manual lookup
AI agent workflow	MCP server with auto-instructions	IDE plugin or CLI

---

Tools

Analysis

Tool	Framework	Purpose
analyze_threat_landscape	STRIDE	Categorise threats from a system description
score_risks	DREAD	Quantitative risk scoring (1-10 per dimension)
assess_privacy_impact	LINDDUN	Privacy threat assessment for personal data
run_pasta_analysis	PASTA	7-stage attack simulation process
build_attack_tree	Attack Trees	AND/OR decomposition of attack paths

Cross-Reference

Tool	Purpose
correlate_frameworks	Map threats across STRIDE → DREAD → LINDDUN → CWE → MITRE ATT&CK
map_to_cwe	Link threats to CWE entries with remediation links
suggest_mitigations	Prioritised mitigation strategies

Documentation

Tool	Purpose
generate_threat_report	Full markdown report combining all frameworks

---

Quick Start

Install

# Using uv (recommended)
uv pip install .

# Or with pip
pip install .

Run the Server

# stdio (default — for IDE integration)
threatprism

# HTTP transport (for shared/team use)
threatprism --transport streamable-http --port 8000

# SSE transport
threatprism --transport sse --port 8000

---

IDE Integration

Cursor

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "threatprism": {
      "command": "threatprism",
      "args": []
    }
  }
}

Or with uv (no install required):

{
  "mcpServers": {
    "threatprism": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/threatprism", "threatprism"]
    }
  }
}

Claude Desktop

Add to claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/, Windows: %APPDATA%\Claude\):

{
  "mcpServers": {
    "threatprism": {
      "command": "threatprism",
      "args": []
    }
  }
}

Or with uv (no install required):

{
  "mcpServers": {
    "threatprism": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/threatprism", "threatprism"]
    }
  }
}

Claude Code (CLI)

claude mcp add threatprism -- threatprism

Or with uv:

claude mcp add threatprism -- uv run --directory /path/to/threatprism threatprism

VS Code (GitHub Copilot)

Add to .vscode/mcp.json:

{
  "servers": {
    "threatprism": {
      "command": "threatprism",
      "args": []
    }
  }
}

Docker

docker build -t threatprism:latest .

{
  "mcpServers": {
    "threatprism": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "threatprism:latest"]
    }
  }
}

---

Agent Workflow

When an AI agent connects, ThreatPrism automatically sends workflow instructions via the MCP handshake. The agent will follow this flow:

┌─────────────────────────────┐
│  1. analyze_threat_landscape │  ← STRIDE categorisation
├─────────────────────────────┤
│  2. score_risks              │  ← DREAD quantitative scoring
├─────────────────────────────┤
│  3. assess_privacy_impact    │  ← LINDDUN privacy analysis
├─────────────────────────────┤
│  4. build_attack_tree        │  ← Attack path decomposition
├─────────────────────────────┤
│  5. correlate_frameworks     │  ← Multi-framework mapping
├─────────────────────────────┤
│  6. generate_threat_report   │  ← Comprehensive documentation
└─────────────────────────────┘

No manual configuration needed — the agent receives the instructions on connect.

---

Example Output

DREAD Score Table

Threat	D	R	E	A	D	Overall	Rating
Authentication Bypass	8.0	7.5	7.0	8.5	6.5	7.5	HIGH
Input Manipulation	9.0	6.0	6.5	7.0	5.5	6.8	HIGH
Session Hijacking	7.5	6.0	5.5	7.0	5.5	6.3	HIGH

Cross-Framework Correlation

Threat	STRIDE	DREAD	LINDDUN	CWE	MITRE
Auth Bypass	SPOO	7.5	IDEN, NON_	CWE-287, CWE-290	T1078, T1110
Data Exposure	INFO	6.8	DISC, LINK, IDEN	CWE-200, CWE-312	T1530, T1567

---

Frameworks

STRIDE (Threat Categorisation)

Classifies threats into six categories: Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, Elevation of Privilege.

DREAD (Risk Scoring)

Quantitative scoring on five dimensions (1-10 each): Damage, Reproducibility, Exploitability, Affected Users, Discoverability. Overall score = average.

LINDDUN (Privacy Threats)

Privacy-specific analysis across seven categories: Linkability, Identifiability, Non-repudiation, Detectability, Disclosure, Unawareness, Non-compliance.

PASTA (Attack Simulation)

Seven-stage process: Business Objectives → Technical Scope → Decomposition → Threat Analysis → Vulnerability Analysis → Attack Modeling → Risk/Impact Analysis.

---

Development

# Install with dev dependencies
uv sync --frozen --all-extras --dev

# Run tests
uv run pytest

# Lint & type-check
uv run ruff check .
uv run pyright

---

License

MIT — see LICENSE.