Skip to content

CLI & Automation

Installation Overview

Code Genius is distributed as a standalone command-line binary that runs locally on your machine.

It does not require a runtime, package manager, or external dependencies.

Supported Operating Systems

Code Genius runs on:

  • macOS (Intel and Apple Silicon)
  • Linux
  • Windows

A platform-specific binary is provided for each supported operating system.

Installation Process

At a high level, installing Code Genius involves:

  1. Downloading the appropriate binary for your operating system
  2. Renaming the binary to code_genius
  3. Adding the binary to your system PATH
  4. Verifying the installation from a terminal

Detailed, platform-specific instructions are available in the Devici platform.

Verifying Installation

After installation, verify that Code Genius is available by running:

code_genius --help

Authentication

Before running Code Genius commands, authenticate with Devici:

./code_genius login

This opens a browser window and securely authenticates your local CLI session.

Analyze Mode

Analyze mode generates a threat model from a single entry point.

Usage

./code_genius analyze [file] [flags]

Where [file] is the path to the file you want to analyze.

Function-Level Analysis

Analyze a specific function:

./code_genius analyze [file] --function [function_name]

List detected functions:

./code_genius analyze [file] --list

Combine both:

./code_genius analyze [file] -lf

Verbosity Levels

Control output detail using the --verbose flag:

./code_genius analyze [file] --verbose 0/1/2
  • 0 — Minimal output
  • 1 — Moderate output (default)
  • 2 — Detailed output

Walk Mode

Walk mode analyzes all supported files in a directory.

./code_genius walk [directory]

Repository Mode (CI/CD)

Repository mode enables non-interactive execution for automation.

./code_genius repo [repository]

This mode requires a Devici API token and does not open a browser.

API Tokens for Code Genius

For CI/CD pipelines and other automated use cases, create an API token in Devici.

  • Code Genius — includes all required permissions

Required Scopes (Manual Configuration)

If you configure scopes manually, ensure the API token includes:

  • cli — write
  • threat-models — write
  • collections — write

Configuration

Configure Code Genius using the config command.

Set Configuration Values

./code_genius config set <key>=<value>

Apply configuration to a profile:

./code_genius config set --profile <profile> <key>=<value>

List configuration options:

List configuration options:

Ignoring Files and Directories .codegeniusignore

Exclude files or directories from analysis:

node_modules
dist
*.log

If no .codegeniusignore file exists, .gitignore is used when available.

Scenario Configuration (codegenius.yaml)

Define multiple threat modeling scenarios:

scenarios:
  - name: ThreatModelOne
    gitignore: false
    include:
      - src/
    exclude:
      - tests/

  - name: ThreatModelTwo
    gitignore: true
    include:
      - api/

Generate a template:

./code_genius walk template

Improving Model Accuracy

Model accuracy is influenced by:

  • Code structure and complexity
  • Framework conventions
  • Inline documentation and annotations
  • Entry point and function selection

Highly abstracted code may require additional refinement.

Known Limitations

  • One repository per run
  • Multi-repository aggregation is not supported
  • Generated models require human review

Next Steps After Generation

After a model is created:

  • Review elements, flows, and attributes
  • Validate inferred trust boundaries
  • Review threats and mitigations
  • Use the Threat Register to manage findings

Code Genius accelerates threat modeling.