NarraNexusNarraNexus
NarraNexus · Getting Started

Quick Start

Three ways to run NarraNexus. Pick the one that fits your situation.


Prerequisites

The Cloud option requires nothing — just a browser. The Desktop app bundles its own runtime, so no separate installs are needed. For Local (from source), make sure you have:

Python 3.13+uv is used for package managementLocal
Node.js 20+Frontend dev serverLocal
tmuxServices run in a single tmux sessionLocal
GitClone the repositoryLocal

On macOS: brew install uv node tmux. On Linux: apt install tmux nodejs and install uv separately.


Full control. Run all services locally via tmux.

1. Clone & Install

git clone https://github.com/user/NarraNexus.git
cd NarraNexus
bash run.sh

The script checks for uv, node, and tmux. If any are missing it prints install instructions and exits. Once dependencies pass, it installs Python and frontend packages, then launches a tmux session with all seven services:

DB Proxy8100SQLite HTTP proxy
Backend8000FastAPI server
MCP7801+Module tool servers
PollerInstance state monitor
JobsScheduled job trigger
BusTriggerMessage bus event trigger
Frontend5173Vite dev server

2. Configure LLM Providers

Open http://localhost:5173 in your browser and log in. On first launch the app redirects you to the Setup page automatically.

Add at least one LLM provider. The fastest path:

  1. 1.Click a preset (NetMind.AI Power, Yunwu, or OpenRouter) and paste your API key
  2. 2.Or add a custom Anthropic / OpenAI provider with your own base URL and key
  3. 3.Or use Claude Code OAuth if you have the CLI installed and logged in

Then assign models to the three required slots:

AgentAnthropicMain dialogue — powers the agent via Claude Agent SDK
EmbeddingOpenAIVector search — generates embeddings for Narratives and Events
Helper LLMOpenAIAuxiliary tasks — topic detection, summarization, entity extraction

Click Apply Changes, then Get Started. You're ready.


What's Next