Prompts Overview

Prompts are pre-configured AI working environments that bundle together skills, documentation, system prompts, and preferences. When activated, a Prompt puts your AI agent into the right "headspace" immediately.

What is a Prompt?

Think of Prompts as AI personality profiles or specialized work modes. Instead of manually configuring your AI for each type of task, a Prompt provides:

• System prompts - Custom instructions for how the AI should behave
• Documentation - Pre-loaded library references (React, TypeScript, etc.)
• Skills - Specialized capabilities and knowledge
• Preferences - Code style, verbosity, response format settings

Key Features

Multiple Active Prompts

You can have multiple prompts active simultaneously. When prompts are stacked, their configurations are merged:

• System prompts are concatenated
• Documentation libraries are combined
• Skills are merged (no duplicates)
• Preferences from later prompts override earlier ones

Prompt Inheritance

Prompts can extend other prompts, inheriting their configuration while adding or overriding specific settings. This allows you to:

• Create a base "TypeScript Developer" prompt
• Extend it with "React + TypeScript" that adds React-specific config
• Further extend with "React Testing" that adds testing tools

Inheritance is limited to 3 levels deep to prevent complexity.

Starter Pack Prompts

Nexus includes 6 pre-configured starter prompts:

Prompt	Description
React + TypeScript Expert	Modern React development with TypeScript, hooks, and best practices
Full-Stack Developer	End-to-end web development across frontend and backend
Testing & QA Engineer	Comprehensive testing strategies and quality assurance
UI/UX Designer	Design systems, accessibility, and user experience
API Developer	REST and GraphQL API design and implementation
TanStack Start + Cloudflare	Full-stack with TanStack Router, Start, and Cloudflare Workers

Getting Started

Using the CLI

# List available prompts
nexus prompt list

# See your active prompts
nexus prompt active

# Activate a prompt
nexus prompt activate react-typescript-expert

# Deactivate a prompt
nexus prompt deactivate react-typescript-expert

# Download a prompt as PROMPT.md
nexus prompt download react-typescript-expert

Using MCP Tools

Your AI agent has access to prompt tools:

• list-prompts - Browse available prompts
• get-prompt - Get a specific prompt by ID
• search-prompts - Search prompts semantically
• save-prompt - Save a new prompt from the agent

Using the Dashboard

Visit the Prompts Dashboard to:

• Browse and search prompts
• Activate/deactivate prompts with one click
• Create custom prompts
• Manage prompt priorities

Auto-Activation

Project Configuration

Create a .nexus/config.json in your project root:

{
  "defaultPrompt": "react-typescript-expert"
}

Prompt Suggestions

The AI can analyze your project and recommend appropriate prompts based on:

• Package.json dependencies
• File structure and patterns
• Technology stack detection

Creating Custom Prompts

You can create your own prompts that combine:

1. System Prompt - Instructions for the AI
2. Libraries - Documentation to pre-load
3. Skills - Capabilities to enable
4. Preferences - Response formatting options

See Creating Prompts for a detailed guide.

Memory Integration

When you save a memory while a prompt is active, the memory is associated with that prompt. This allows for:

• Prompt-specific knowledge accumulation
• Context-aware memory recall
• Better organization of project learnings

Next Steps

• CLI Commands - Full CLI reference for prompts
• MCP Tools - Using prompts with AI agents
• Creating Prompts - Build your own custom prompts
• Best Practices - Tips for effective prompt usage