Prompt Smith

A prompt engineering solution to manage Gen AI prompts easily.

Features

• Self-hosted option with full control over data
• Dockerized for easy deployment
• RESTful API for easy integration
  • With SDK for Python and Node.js.
• API Key management through centralized UI
• Prompt Management through centralized UI
  • Variants
  • Versioning (database level)

• PromptSmith Intro This is a video to demonstrate the features of Prompt Smith

How to use

1. Start the service locally docker-compose up --build
2. Access the at http://localhost:3000
3. Default login credentials (this could be set in docker compose file, please do not use default credentials in production)
   • email: admin@localhost.lan
   • password: AwesomePromptsManagement
4. Create a prompt at http://localhost:3000/prompt
5. Create an API key at http://localhost:3000/key
6. Get your prompts with ease through RESTful API

curl "http://localhost:3000/api/bff/api/sdk/prompt/{prompt-key}" \
     -H 'Authorization: Bearer {api-key}' \
     -H 'Content-Type: application/json; charset=utf-8' 

6. Or Get prompts through SDK
   • Python SDK and Python Example
   • Node.js SDK and JS/TS Example

We have deployed a Demo Instance

• Please note the data may not be persisted and will be deleted scheduledly.

Development and Contribution

• Contribution Guide

Features in the future

These features were not planned in the beginning but could be added in the future.

• [ ] Prompt Compression, to reduce the cost of token usage but still maintain the quality of prompts
• [ ] SaaS version of Prompt Smith.
• [ ] Trace of prompt usage.
• [ ] Webhook support.
• [ ] Prompt Management through CLI.
• [ ] Integration with analytics tools, such as Google Analytics, Mixpanel, etc.

Please feel free to open an issue or PR for any feature requests or bug reports.