3 ms·
Show HN: Open-source AI prompt engineering workbench with systematic evaluation
PromptForge: Open-source AI prompt engineering workbench with systematic evaluation
I built PromptForge after getting frustrated with the manual trial-and-error approach to prompt engineering. Most tools are just text editors, but I wanted something that brings software engineering discipline to prompt development.
Key technical features:
- AI-assisted prompt generation (stop starting from scratch)
- Systematic evaluation engine that auto-generates test suites
- Multi-model support (Claude, GPT-4, Azure OpenAI)
- Go backend with SQLite for performance and simplicity
- One-line Docker deployment
The evaluation system automatically tests for robustness, safety, factual accuracy, and creativity across different scenarios. Think unit testing but for prompts.
Currently #1 on Product Hunt today after 5+ hours, which validates that developers need better prompt engineering tools.
Open source (GPLv3): https://github.com/insaaniManav/prompt-forge https://github.com/insaaniManav/prompt-forge
Demo: https://demo.arcade.software/bFGTYb7AuRV33Kei7ZFQ https://demo.arcade.software/bFGTYb7AuRV33Kei7ZFQ
PH URL : https://www.producthunt.com/products/promptforge-2 https://www.producthunt.com/products/promptforge-2
Would love feedback from the HN community, especially on the evaluation methodology and technical approach.