Getting Started

Installation, setup, and your first steps with the AI Prompt Library.

---

System Requirements

Requirement	Details
Node.js	18 or later (for CLI and desktop builds)
npm dependencies	None — zero external packages
Browser	Any modern browser (for Prompt Workshop)
Disk space	~2 MB (the entire library)

Don't have Node.js? Download it from nodejs.org — grab the LTS version. Or skip Node entirely and use the browser-only option.

---

Installation Options

Option 1: Clone and Run (No Install Needed)

The fastest way to start. Clone the repository and run the CLI directly:

git clone https://github.com/diShine-digital-agency/ai-prompt-library.git
cd ai-prompt-library
node bin/prompt-lib.js --help

That's it. No npm install, no dependencies, nothing to set up.

Option 2: Install Globally via npm

Install once and use the prompt-lib command from anywhere:

npm install -g @dishine/prompt-library
prompt-lib --help

This registers the prompt-lib command system-wide.

Option 3: Link for Development

If you cloned the repo and want the prompt-lib command available globally while you work on the code:

cd ai-prompt-library
npm link
prompt-lib --help

Changes to source files take effect immediately — no rebuild needed.

Option 4: Browser Only (No Node.js Needed)

Just open viewer.html in any browser. That's the Prompt Workshop — a standalone file with everything built in:

• No server required
• No internet required
• No Node.js required
• All 82+ prompts embedded in the file

Simply double-click viewer.html or drag it into your browser.

Option 5: Desktop App (Build from Source)

Build native desktop apps for macOS, Linux, or Windows:

git clone https://github.com/diShine-digital-agency/ai-prompt-library.git
cd ai-prompt-library
./desktop/build-all.sh

Output goes to dist/. See the Desktop Apps page for platform-specific guides.

Platform	Build Script	Requirements
macOS	./desktop/build-macos.sh	Xcode CLI Tools for native app
Linux	./desktop/build-linux.sh	Python 3, GTK, WebKitGTK
Windows	./desktop/build-all.sh	Git Bash or WSL
All	./desktop/build-all.sh	Node.js 18+, Bash

---

First Commands to Try

Once installed, try these commands to explore the library:

# List all 82+ prompts grouped by category
prompt-lib list

# Search for prompts by keyword
prompt-lib search "chain of thought"

# Show full details of a specific prompt
prompt-lib show chain-of-thought

# See all categories with counts
prompt-lib categories

# Get library statistics
prompt-lib stats

# Show a random prompt for inspiration
prompt-lib random

---

Opening the Prompt Workshop

The Prompt Workshop is a browser-based visual tool with 7 tabs. Open it from the CLI:

prompt-lib viewer

Or open viewer.html directly in any browser — no server needed.

The Workshop provides everything the CLI does, plus:

• Visual search and filtering
• AI Playground (send prompts to GPT/Claude/Gemini)
• Dark/light mode
• My Library (save, edit, export/import prompts)
• Keyboard shortcuts

---

Quick Tutorial: Finding and Using a Prompt

Step 1: Find a Prompt

prompt-lib search "code review"

Output shows matching prompts ranked by relevance:

Search results for "code review" (3 found)

  code-review (score: 200)
    Code Review Checklist [intermediate] in development
    tags: code-review, quality, checklist

  code-refactoring-review (score: 110)
    Code Refactoring Review [intermediate] in development
    tags: refactoring, review, patterns

Step 2: View the Full Prompt

prompt-lib show code-review

This displays the complete prompt content including the template, tips, and examples.

Step 3: Build It Interactively

prompt-lib use code-review

The CLI detects {{placeholders}} in the template and asks you to fill each one:

Building prompt: Code Review Checklist
3 field(s) to fill in

  programming language: Python
  code snippet: def calculate_total(items): ...
  focus area: security and error handling

The filled prompt is automatically copied to your clipboard — ready to paste into any AI model.

Step 4: Compose a Multi-Layer Prompt

prompt-lib compose

This lets you combine:

1. System prompt (persona and rules) — e.g., "Coding Assistant"
2. Framework (reasoning technique) — e.g., "Chain-of-Thought"
3. Task template (domain-specific) — e.g., "Code Review"

The result is a powerful layered prompt that outperforms any single template.

---

What's Next?

Want to...	Go to...
Learn all CLI commands	CLI Reference
Explore the browser tool	Prompt Workshop
Master prompting techniques	Prompting Techniques
Compare AI models	AI Models Guide
Use the quality tools	Tools: Linter, Optimizer, Recommender
Build desktop apps	Desktop Apps
Contribute	Contributing

---

Navigation: ← Home  |  CLI Reference →