promptr

Lightweight Prompt Engineering CLI for developers. Improve prompts with offline rules or optional LLM mode.

  ██████╗ ██████╗  ██████╗ ███╗   ███╗██████╗ ████████╗██████╗ 
  ██╔══██╗██╔══██╗██╔═══██╗████╗ ████║██╔══██╗╚══██╔══╝██╔══██╗
  ██████╔╝██████╔╝██║   ██║██╔████╔██║██████╔╝   ██║   ██████╔╝
  ██╔═══╝ ██╔══██╗██║   ██║██║╚██╔╝██║██╔═══╝    ██║   ██╔══██╗
  ██║     ██║  ██║╚██████╔╝██║ ╚═╝ ██║██║        ██║   ██║  ██║
  ╚═╝     ╚═╝  ╚═╝ ╚═════╝ ╚═╝     ╚═╝╚═╝        ╚═╝   ╚═╝  ╚═╝

What is promptr?

promptr is an offline-first CLI that analyzes and improves prompts using practical prompt-engineering skills.

• Fast, local, and deterministic in offline mode
• Deeper rewriting in --llm mode (Anthropic)
• Clear before/after scoring and skill-level changes

Installation

npm install -g promptr-cli

Local development:

npm install
npm link
promptr skills

Quick Start

# show skills
promptr skills

# offline auto-improve
promptr improve ./prompt.txt --auto

# offline auto-improve + save
promptr improve ./prompt.txt --auto --save

# manual skill selection
promptr improve ./prompt.txt --skills clarity,output-format,role-injection

# llm mode
promptr improve ./prompt.txt --auto --llm

CLI Reference

promptr skills

Lists all available skills.

promptr skills

promptr improve <file>

Improves a prompt file.

promptr improve <file> [options]

Options:

• --auto: auto-detect needed skills
• --skills <comma-separated>: apply selected skills only
• --save: save output as <input>.improved.txt
• --llm: use Anthropic-powered mode

Modes

Offline mode (default)

Best for speed, privacy, and zero API cost.

promptr improve ./prompt.txt --auto

LLM mode (--llm)

Best for nuanced prompts and deeper rewrites.

promptr improve ./prompt.txt --auto --llm

Use Both Modes Together (Recommended)

# Step 1: quick cleanup offline
promptr improve ./prompt.txt --auto --save

# Step 2: deeper refinement with llm
promptr improve ./prompt.improved.txt --auto --llm --save

LLM Setup

Set your Anthropic API key:

# PowerShell
$env:ANTHROPIC_API_KEY="your_api_key"

# bash/zsh
export ANTHROPIC_API_KEY="your_api_key"

Built With

• Offline engine: custom rule-based prompt optimization
• LLM mode: Anthropic Claude via @anthropic-ai/sdk

Skills

| Skill | Description | |---|---| | clarity | Remove vague instructions and improve precision | | role-injection | Add a strong role/persona | | output-format | Enforce explicit output structure | | chain-of-thought | Add reasoning guidance for complex tasks | | constraint-guard | Detect contradictory instructions | | token-trim | Remove filler and redundant wording |

Example

Input:

be helpful and summarize this article and so on. do your best.

Run:

promptr improve ./sample-prompt.txt --auto --save

Expected improvements:

• Adds role definition
• Rewrites vague phrases
• Adds clear output format
• Raises prompt quality score

Author

Mohamed Hamed

Built for developers who want fast, practical prompt engineering workflows.