SMEEware Chat runs where you do.
Watch it think, use tools and answer in real time. Bring your files, your voice and whichever model you like. Nothing leaves your machine unless you send it.
~/smeeware-chatgit log --graph
Every feature is a commit. Scroll down and watch the history play back.
What it does
Three things you will reach for every day.
Reasoning, tool calls and the reply stream in live. Keep typing while it works and your note joins at the next step.
How it worksA hands-free call with a free neural voice. Start talking whenever you want to cut in.
How it worksPoint a workspace at a folder. Files, search and your tools come along with every question.
How it worksLocal or hosted. You decide per message.
Switch models in the middle of a chat. Nothing else changes.
On your machine
Ollama models run right next to you. Nothing leaves the box.
In the cloud
DeepSeek, OpenAI and friends when you need more muscle. Your keys, your bill.
Encrypted on disk
Your chats are locked with a key made from your password. That key only ever lives in memory.
- title
- Release checklist
- message
- What's still open before the release?
- key
- only in memory
Running in two minutes
You need Python 3.10+ and Node 20+. The script takes care of the rest.
curl -N http://127.0.0.1:8000/api/v1/chat/stream \
-H "Content-Type: application/json" \
-d '{
"messages": [{"role": "user", "content": "Hello!"}],
"model": "deepseek-v4-flash"
}'Going online?
Local stays keyless. Once it is on the internet, every request needs a key. You see it once and send it as a bearer token.
Shown once. Keep it somewhere safe.
The whole API
16 endpoints. That's all there is to learn.
Ready when you are.
Everything above is one command away.