Local-first by design
Your model, index, history, and source code remain on your machine.
Local-first AI for developers
Build with AI without sending your code to the cloud.
Most chat agents struggle with Ollama and small models. OwA is built specifically for developers running practical local hardware—with focused tools that help 4–9B models work clearly inside a real repository.
Your LLM runs on your machine. Your code stays on your machine. OwA works within your workspace—nothing leaves your hardware.
- ✓ No cloud
- ✓ No telemetry
- ✓ No API key
$ owa index .
scanning workspace…
✓ 1,284 files indexed locally
$ ask "where is auth handled?"_
MODELornith:9b
⌾PrivateNothing leaves
Why OwA exists
Built for local hardware, private workflows, and real coding tasks.
Most coding agents assume a giant cloud model. OwA starts with the hardware developers already own, then makes focused local models genuinely useful.
ornith:9bqwen2.5-codernomic-embed-text
Small model friendly
Clear tools and compact context help 4–9B models do focused, useful work.
More control, less leakage
Workspace boundaries and explicit confirmations keep actions inspectable.
Inside the workspace
Everything OwA needs.
Nothing it doesn’t.
Focused tools for the work developers actually do, connected to the model running on their machine.
Semantic code search
Index the project locally and find the files that matter by meaning, not filename.
› owa index ./workspaceFile reading & patching
Read, edit, and patch targeted locations while keeping every change reviewable.
› owa patch --confirmGit awareness
Inspect status, diffs, and recent history without leaving your repository.
› owa ask "what changed?"Review & safety checks
Run structured Python checks and confirm shell actions before they execute.
› owa review app/Under the hood
Transparent all the way down.
A lightweight Python app, direct Ollama integration, local indexing, and a compact tool system you can understand.
Python 3.11+OllamaSQLiteFastAPIUvicornRichrequestshttpx
Keep OwA building
Open source grows with support.
If OwA helps your local workflow, sponsorship keeps the project maintained, documented, and improving.