PromptMesh

What is PromptMesh and what is it for?

PromptMesh is a Holochain app for sharing AI prompts, where peers add suggestions to each other's prompts and each node relays the combined result to the language model it is using. It exists so a prompt can be improved by a group without a central service holding the prompts or the models.

PromptMesh is an open-source project on GitHub that relays AI prompts between peers. It is built on Holochain, a framework in which each person runs their own copy of an app and those copies share data directly rather than through one company's servers. The repository introduces it in one line as a peer-to-peer AI prompt relay on Holochain.

The idea is to improve a prompt the way a group improves a document. One peer writes a prompt and posts it; other peers add suggestions and improvements, which the app keeps linked to the original. Each node then gathers the suggestions, sends the combined prompt to the language model it has chosen, and posts the refined result back for the others. Every node uses its own model connection, so the interface can work with a model running on your own machine.

A practical starting point is the repository's UI-only route, which works on any operating system: install Ollama, pull the llama3 model, start it so the browser can reach it, then open ui/index.html — the interface falls back to demo mode, so the relay and the model connection work before Holochain itself is running. The README then walks through the full peer-to-peer setup with Nix and the hc command-line tools, for Linux and macOS or for Windows through WSL2. The code and current setup notes live at https://github.com/ta10101/promptmesh.

From 1 public source, checked 23 September 2026.