NewServer

LuminaryChat

A minimal, open-source web chat server for AI with a clean, fast foundation and simple extensibility.

Overview

LuminaryChat is a lightweight, welcoming OpenAI-compatible API server that brings historical personas to life. Point any OpenAI-compatible chat client at it, and it injects well-crafted, historically grounded system prompts so responses stay in-character.

Highlights

Five historical personas included out of the box:

  • luminary/confucius — duty, ritual, self-cultivation; redirects grievance to conduct.
  • luminary/leonardo_da_vinci — observation and experiment; bridges art, anatomy, engineering, and nature.
  • luminary/marie_curie — evidence and rigor; demands data, controls, and measurement.
  • luminary/socrates — questions over answers; examines definitions and contradictions.
  • luminary/sun_tzu — strategy as terrain; deception, positioning, and winning without fighting.
  • OpenAI-compatible endpoints: /v1/models and /v1/chat/completions.
  • Streaming support: Server-Sent Events with chunk rewriting to match model/persona ID.
  • Sensible defaults: .env config, structured JSON logs, health and optional metrics endpoints.

Features

Lightweight

Minimal server footprint focused on fast, responsive interactions.

OpenAI-Compatible Endpoints

Implements /v1/models and /v1/chat/completions for easy client integration.

Streaming Responses

Server-Sent Events with chunk rewriting to match persona IDs; smooth token streaming.

Sensible Defaults

.env configuration, structured JSON logs, health and optional metrics endpoints.

Connect from Terminal

Prefer the terminal? Use chaTTY to connect to a LuminaryChat server using an OpenAI-compatible endpoint.