Prompt-run – run .prompt files against any LLM from the terminal
Git-friendly `.prompt` format with `prompt diff` side-by-side output beats scattered Notion docs.

Beautiful node-based UI for prompt branching, but prompt iteration tools already exist everywhere.
Prompt engineers, AI researchers, teams experimenting with complex LLM workflows
LangSmith · Promptfoo · LM Studio
I built OXPT because I was tired of the "Linear Chat Prison" of current LLM interfaces. Whenever I try to implement complex logic like Self-Reflection or Tree-of-Thought, I lose track of which branch worked best.
It’s a visual playground where you can:
Branch your thoughts: See multiple AI responses side-by-side on a non-linear canvas.
Version control for prompts: Save and compare every iteration instead of scrolling through endless chat history.
Modular Logic: Treat prompts like building blocks.
I wanted a more systematic way to manage my own prompt experiments.
I' d love some brutal technical feedback from this community. What do you think about the node-based UI compared to traditional list-based chats?
(P.S. The logo was inspired by a drawing my son made for me!)
https://www.oxpt.online/assets/images/scr_01.png
Git-friendly `.prompt` format with `prompt diff` side-by-side output beats scattered Notion docs.
Prompt versioning and rollback without redeploy — but PromptLayer and Langfuse already own this.
Prompt versioning is nice, but web tools and Cursor already do side-by-side comparison.
Git-style prompt versioning in a single SQLite file—no SaaS, no telemetry.
Static scanner catches prompt injections in code before runtime, unlike runtime guards.
GitHub for prompts is an interesting bet, but PromptBase and FlowGPT already exist.