I've been using coding agents like Claude Code, Codex, and OpenCode for a while now. They've genuinely changed how I work — but almost all of them live in the terminal.
I really like the openness and model independence of tools like OpenCode. What I wanted was that same flexibility inside something that feels like a real desktop application: install it, add an API key, open a project, and start working.
So I built Velloc Code — an open-source, Windows-first desktop coding agent.
Why a desktop UI?
Terminal/TUI agents are great, and I'm not trying to replace them. The argument for a desktop app is simple:
- No environment setup — install it, add an API key, open a project folder. No shell config, no dependency bootstrapping.
- A UI you can reason about — sessions, project memory, and agent configuration are visible and inspectable, not hidden in config files.
- One app for your whole workflow — reading code, running agents, reviewing output, and checking what the agent actually did all live in the same window.
The philosophy is: make the default experience easy, and don't lock anyone in.
Multi-model workflows
The core idea is that one model doesn't have to do everything.
You can start with a single model and a single agent. When you outgrow that, you can assign different models to different agents or tasks:
- a cheaper/faster model for research and planning
- a stronger model for implementation
- another model for review and testing
The same project can split a larger task across agents, each using the model that fits its role — without changing your workflow or your provider.
It's provider-agnostic: bring your own providers or custom model endpoints, and switch models without touching your setup.
What else it does
- Project memory across sessions — context survives, so you're not re-explaining your project every time.
- Extensible — skills and Claude Code-compatible plugins.
- Inspectable — you can inspect and customize the agent itself, instead of treating it as a black box.
The direction
The openness and model independence of open-source coding agents, with a polished desktop UI and a much easier starting point.
It's Windows-first right now and still early.
I'd really value feedback from people already using Claude Code, Codex, OpenCode, Cursor, or similar:
- Would you actually prefer a desktop UI, or is terminal/TUI still the ideal interface?
- Would you use different models for different agents or tasks?
- What would make you switch from your current coding agent?
- What feels unnecessary or over-engineered?
Check it out: velloc.app — and the source is on GitHub.
I'm the developer, so feel free to be critical — that's mainly why I'm posting it here.
Top comments (0)