I built Gitizi an open-source library of executable prompts. Since LLM workflows start with executing a prompt, I wanted to build something around this basic unit for generative AI.
Like a simplified version of Github for prompts.
I checked different prompt libraries but they're limited to copy-paste. So I built a platform where prompts can be run and orchestrated.
I think it's handy because different LLMs perform differently on different tasks - users can chain workflows across multiple models.
I created a simple markup language based on Laravel Blade to define prompt blocks and how they're executed.
I'm planning to implement MCP tooling in the workflow but don't want to overbuild without external feedback.
Would love feedback on the UX and what features would make this useful for your workflows.
Happy to answer questions!