Skip to main content
Manage projects, requirements, and tasks from your terminal. Common use cases:
  • Fetch a requirement’s tasks - Get the tasks for a requirement for your coding agent to implement.
  • Save agent plans - Save your coding agent’s plans to share and implement later.
  • Refine vague ideas - Transform rough concepts into clear, actionable requirements with AI assistance.

​
Installation

Install the BrainGrid CLI globally using npm: 
npm install -g @braingrid/cli

​
Authentication

Before using the CLI, authenticate with your BrainGrid account: 
braingrid login

​
Quick Start: One-Minute Flow

Get started with BrainGrid in four simple steps:
1

Initialize

Initialize your project with BrainGrid:
braingrid init
2

Specify

Create a requirement with AI refinement:
braingrid specify --prompt "Add user authentication with OAuth2"
3

Build

Get complete implementation plan:
braingrid requirement build REQ-1
The CLI automatically detects your current project from .braingrid/project.json and can extract requirement IDs from your git branch names (e.g., feature/REQ-123-auth).

​
AI Coding Integration

BrainGrid CLI integrates seamlessly with AI coding tools:

Working with Claude Code

Skills and slash commands for Claude Code.

Working with Cursor

Rules and slash commands for Cursor.

​
Typical Workflow

Here’s how BrainGrid CLI fits into your development process:
  1. Capture Ideas - Use braingrid specify to refine vague ideas into structured requirements
  2. Build Features - Create a git branch and let your AI agent work through tasks
  3. Track Progress - Update task status with braingrid task update as you complete work

​
Auto-Detection Features

The CLI includes smart auto-detection to minimize typing:
  • Project Detection - Automatically finds project from .braingrid/project.json
  • Branch-Based Requirements - Extracts requirement IDs from branch names like feature/REQ-123-description
  • Flexible ID Formats - Accepts REQ-123, req-123, 123, or full UUIDs

​
Output Formats

Commands support multiple output formats for different use cases:
  • table (default) - Human-readable tables for terminal viewing
  • json - Machine-readable format for scripting and automation
  • xml - Structured data for enterprise integrations
  • markdown - Rich formatted output perfect for AI agents
Use the --format flag on any command: 
braingrid requirement list --format json
braingrid requirement build REQ-123 --format markdown

​
Next Steps

Command Reference

Explore all available commands and their options

MCP Server

Set up BrainGrid as a Model Context Protocol server