⚠️ Legacy — frozen at v2.4.0, no longer maintained. The prompt-lib CLI documented here still works against the prompts/ library, but receives no new features or bug fixes. For the maintained Prompt Workshop — with Live Battle & Leaderboard, private accounts & chat, AI chat with files and memory, and many more models — use prompt.dishine.it.

Complete documentation for every prompt-lib command (legacy, v2.4.0).

The prompt-lib CLI is the legacy terminal interface to the AI Prompt Library. It provides commands for discovering, building, composing, linting, optimizing, and managing prompts — all with zero npm dependencies.

prompt-lib <command> [arguments]

Global install:

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

Local (no install):

node bin/prompt-lib.js --help

Lists all prompts grouped by category, showing slug, title, and difficulty level.

prompt-lib list

Output:

Prompt Library (82 prompts)

business (12)
  client-communication - Client Communication Template [intermediate]
  competitive-intelligence - Competitive Intelligence Brief [advanced]
  ...

data (10)
  dashboard-spec - Dashboard Specification [intermediate]
  ...

Searches prompts by keyword across titles, tags, categories, and content. Results are ranked by relevance score.

prompt-lib search <query>

Scoring system:

Multi-word queries score each word independently. A prompt matching "code" and "review" in the title scores 200 points.

Examples:

prompt-lib search "chain of thought"
prompt-lib search "marketing email"
prompt-lib search "code review security"
prompt-lib search "claude xml"
prompt-lib search "data pipeline"

Displays the complete content of a prompt including metadata, template, tips, and examples.

prompt-lib show <slug>

Example:

prompt-lib show chain-of-thought
prompt-lib show code-review
prompt-lib show landing-page-copy

The output includes:

• Title, category, difficulty, tags, compatible models
• Full prompt content with all sections
• File path within the library

Extracts the template section from a prompt, detects {{placeholders}}, and prompts you to fill each one. The completed prompt is copied to your clipboard.

prompt-lib use <slug>

Example:

prompt-lib use code-review

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

  programming language: Python
  code snippet: def calculate_total(items): return sum(i.price for i in items)
  focus area: security and performance

──────────────────────────────────────────────────────────
  YOUR PROMPT (ready to paste)
──────────────────────────────────────────────────────────

[Filled prompt content here]

  Copied to clipboard.

Extracts and copies the template section of a prompt directly to clipboard, without interactive filling.

prompt-lib copy <slug>

Example:

prompt-lib copy chain-of-thought
prompt-lib copy sql-query-builder

Interactive command that lets you layer a system prompt + framework + task template into a single powerful prompt. Placeholders across all layers are filled interactively.

prompt-lib compose

Workflow:

1. Pick a system prompt (e.g., Coding Assistant, Research Assistant)
2. Pick a reasoning framework (e.g., Chain-of-Thought, Few-Shot Patterns)
3. Pick a domain template (e.g., Code Review, SQL Query Builder)
4. Fill any {{placeholders}} across all three layers
5. Get the composed prompt copied to clipboard
6. Optionally save the composition for later

Example output:

# SYSTEM PROMPT
You are a senior software engineer...

# REASONING FRAMEWORK
Think through this problem step by step...

# TASK TEMPLATE
Review the following code for...

Saved compositions go to ~/.prompt-library/saved-prompts.json.

Interactive command to build a new system prompt with custom dynamic fields.

prompt-lib create

Workflow:

1. Enter a title, category, tags, difficulty, and target models
2. Define dynamic fields (become {{field_name}} placeholders)
3. Write the prompt body using your fields
4. Prompt is saved to ~/.prompt-library/custom-prompts.json
5. Optionally fill and use it immediately

Example:

CREATE — build a new system prompt with custom fields

  Prompt title: Technical Blog Writer
  Category: custom
  Tags: blog, technical, writing
  Difficulty: intermediate

  Field name: topic
    Description: The topic of the blog post
  Field name: audience
    Description: Target audience level

  > Write a technical blog post about {{topic}} for {{audience}}...

Your custom prompts appear in prompt-lib list and prompt-lib search alongside built-in prompts.

Interactive command that guides you through creating a prompt using one of 5 built-in frameworks.

prompt-lib generate

Available frameworks:

Each framework asks context-specific questions and generates a production-ready prompt. You can save the result as a custom prompt.

Analyzes any prompt against 14 quality rules, producing a score (0–100), letter grade (A–F), and prioritized suggestions.

prompt-lib lint

Paste your prompt (multi-line, end with two empty lines), and get:

LINT RESULTS

  Score: 72/100 (Grade: C)
  Rules: 10/14 passed | 89 words

  ✅ Passing:
     • Role definition
     • Clear task
     • Constraints or rules
     ...

  💡 Suggestions to improve:
     → Specify the output format: "Respond in [format]"
     → Add examples of expected input/output
     ...

See Tools: Linter, Optimizer, Recommender for the full rule list and scoring details.

Analyzes and rewrites your prompt using content-aware optimization. Works entirely offline.

prompt-lib optimize

The optimizer:

1. Detects the domain (coding, writing, marketing, data, business, education, image)
2. Removes filler words and excessive politeness
3. Strengthens weak verbs
4. Replaces vague language with specific terms
5. Adds domain-specific structure (role, constraints, output format)
6. Shows before/after scores and all changes made

Score: 35 → 88 (+53)
Changes made:
  • Removed filler words and redundant phrases
  • Replaced weak/hedging language with direct instructions
  • Added domain-specific role (coding)
  • Organized prompt into clear labeled sections
  • Added coding-specific output format

Analyzes your use-case description and suggests the best prompts, including an optimal system prompt + framework + template combination.

prompt-lib recommend <query>

Example:

prompt-lib recommend "I need to write a landing page for a SaaS product"

RECOMMENDATIONS for: "I need to write a landing page for a SaaS product"

  Suggested combination:

    🧠 System prompt:  Content Writer (content-writer)
    🔧 Framework:      Few-Shot Patterns (few-shot-patterns)
    📝 Template:       Landing Page Copy (landing-page-copy)

  Top matching prompts:

    Landing Page Copy
      slug: landing-page-copy | category: marketing | score: 65
    ...

Lists all saved compositions and custom prompts from ~/.prompt-library/.

prompt-lib saved

Opens the Prompt Workshop (visual browser tool) with all prompts loaded.

prompt-lib viewer

The command injects current prompt data into viewer.html and opens it in your default browser.

Shows all prompt categories with their prompt counts.

prompt-lib categories

Displays a random prompt from the library — great for inspiration.

prompt-lib random

Shows comprehensive statistics: total prompts, categories, unique tags, models covered, difficulty distribution, and top tags.

prompt-lib stats

prompt-lib --help

prompt-lib --version

The copy, use, and compose commands automatically copy results to your clipboard. Platform-specific tools used:

If no clipboard tool is available, the CLI shows the prompt content and asks you to copy manually.

prompt-lib search "sql"          # Find SQL-related prompts
prompt-lib show sql-query-builder # Read the full prompt
prompt-lib use sql-query-builder  # Fill placeholders and copy

prompt-lib compose               # Layer system + framework + template
# or for AI-powered suggestions:
prompt-lib recommend "build a REST API" # Get recommendations first

prompt-lib lint                  # Score your prompt
prompt-lib optimize              # Rewrite with best practices

prompt-lib create                # Build a custom prompt
prompt-lib saved                 # View your saved prompts
prompt-lib use my-custom-prompt  # Use it anytime

Navigation: ← Getting Started  |  Prompt Workshop →