lucinate
lucinate is a terminal-native AI chat client that connects to OpenClaw agents, Hermes profiles, or any OpenAI-compatible endpoint. It solves the problem of managing multiple AI backends and agents without leaving the terminal, offering streaming responses and markdown rendering. It is for developers who prefer a keyboard-driven, mouse-free chat interface.
✨ Key features
- Multiple backends: OpenClaw, Hermes, OpenAI-compatible endpoints
- Streaming responses with conversation history and multi-agent support
- Create and delete agents directly from the TUI
- Markdown rendering for assistant messages
- Tool call cards showing execution status (OpenClaw)
- Local and remote shell commands with ! and !!
🎯 Use cases
- Chat with OpenClaw agents from the terminal
- Interact with local LLMs via Ollama or LM Studio
- Run one-shot queries with lucinate send for scripting
- Manage cron jobs and routines without leaving the TUI
- Switch between multiple AI backends and agents seamlessly
📦 Installation
🧰 Requirements: Requires a terminal on macOS, Linux, or Windows; for OpenClaw, a gateway with device pairing; for OpenAI-compatible, an endpoint URL; for Hermes, a profile server. No API keys mentioned.
Install
With Homebrew (macOS / Linux):
brew trust --formula lucinate-ai/tap/lucinate
brew install lucinate-ai/tap/lucinate
Or via the install script (macOS / Linux):
curl -fsSL https://raw.githubusercontent.com/lucinate-ai/lucinate/main/install/lucinate.sh | sh
On Windows, in PowerShell:
irm https://raw.githubusercontent.com/lucinate-ai/lucinate/main/install/lucinate.ps1 | iex
Or, if you have Go 1.25+:
go install github.com/lucinate-ai/lucinate@latest
Or build from source:
git clone https://github.com/lucinate-ai/lucinate.git
cd lucinate
go build -o lucinate .
🚀 Usage
2. Connect
lucinate
On first launch, a Connections picker appears to add a backend. For OpenClaw, pair with the gateway:
openclaw device list --pending
openclaw device approve <device-id>
Then select an agent and start chatting. For one-shot mode:
lucinate send --connection my-con --agent main "summarise this PR description"
❓ FAQ
How do I connect to an OpenAI-compatible endpoint like Ollama?
In the Connections picker, choose 'OpenAI-compatible' and enter the base URL ending in /v1. lucinate will use the /v1/chat/completions endpoint.
Can I run shell commands from the chat?
Yes, prefix with ! for local commands and !! for remote commands on the gateway. Remote commands are subject to the gateway's exec security policy.
How do I create a routine?
Use /routines to manage routines. Each routine is a list of steps stored in ~/.lucinate/routines/<name>/STEPS.md. You can set mode to auto or manual.
What is the difference between lucinate send and lucinate ask?
send requires explicit --connection and --agent flags, while ask uses defaults configured in /settings under Ask command defaults, so you can omit them.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.