PromptMage is a Python framework designed to simplify the development of complex, multi-step applications based on large language models (LLMs). It provides an intuitive interface to create, test, compare, and manage LLM workflows as a self-hosted solution, with built-in tooling for prompt versioning, testing, and easy integration.
Overview
PromptMage focuses on making prompts first-class citizens with version control and collaboration features, enabling manual and automatic testing, and providing an API for easy deployment. It aims to bridge the gap in LLM workflow management by offering pragmatic, self-hosted capabilities for developers, researchers, and organizations.
How to Use PromptMage
- Install and set up PromptMage locally or on your server with quick installation steps.
- Explore the Prompt Playground to test, compare, and refine prompts in an intuitive interface designed for rapid iteration.
- Leverage the auto-generated API powered by FastAPI for easy integration and deployment.
- Manage prompts with version control to track changes, collaborate, and reuse prompts across projects.
Key Features
- Prompt Playground for fast iteration and rapid testing
- Prompts as first-class citizens with integrated version control
- Manual and automatic testing/validation of prompts
- Built-in, automatically created API using FastAPI for easy integration
- Type-hint everything for automatic inference and validation
- Easy sharing of results with domain experts and stakeholders
- Self-hosted workflow management to bridge the LLM gap
- Evaluation Mode to assess prompt performance before deployment
Core Concepts
- Prompts and prompts workflows are versioned and collaboratively developed
- Automatic API generation simplifies integration into applications
- Validation and testing workflows help ensure reliability prior to deployment
- Self-hosted design emphasizes data control and privacy
Roadmap & Philosophy
- The project emphasizes fast iteration, prompt-centric workflows, and easy integration.
- Ongoing enhancements are planned, including more features and enhancements as the project evolves.
Projects & Development
- Example project: product-review-research – an AI web app built with PromptMage to analyze trustworthy online reviews.
- Development and contribution guidelines are documented to help contributors participate effectively.
How to Contribute
- Report bugs with detailed descriptions and reproduction steps.
- Improve documentation and propose enhancements via issues or pull requests.
- Follow the contribution guidelines and ensure alignment with project goals.
- Create issues before submitting PRs to discuss changes.
- Adhere to coding standards; check DEVELOPMENT.md for details.
- Contact: [email protected]
Safety & Use Considerations
- The API is designed for self-hosted deployments to maintain control over data and workflows.
Core Features Summary
- Prompt Playground for rapid testing and iteration
- Prompts with version control and collaboration features
- Manual and automatic prompt testing/validation
- Auto-generated FastAPI-based API for easy integration
- Type hints for automatic inference and validation
- Easy sharing of results with stakeholders
- Self-hosted workflow management for LLM applications