AI vibe coding
with visual feedback
Local-first desktop app that wraps Claude Code with a rich visual interface — tool cards, run history, multi-provider switching, and more.
All data stored locally. No cloud backend.
The CLI is powerful.
But it's still a terminal.
Claude Code and similar AI coding CLIs are incredible tools. But running in a terminal means missing out on a lot.
No persistent dashboard
Close the terminal, lose your context. No way to see all sessions at a glance.
No visual diff review
Code changes appear as raw text. No syntax highlighting, no inline comparison.
No cross-session history
Past conversations disappear. No way to search, replay, or fork previous work.
Features
Everything the terminal can't provide
Persistent dashboard, visual diffs, cross-session history, multi-provider switching — and more.
Visual Tool Cards
Every tool call (Read, Edit, Bash, Grep, Write) rendered as an inline card with syntax-highlighted diffs and structured output. One-click copy, expandable details.
Multi-Provider Switching
Use Claude Code with 15+ API providers — DeepSeek, Kimi, Zhipu, MiniMax, OpenRouter, Ollama, and more. Hot-switch without restarting.
Run History & Replay
Browse all past sessions, full event replay, resume or fork from any point. Import CLI sessions with one click. Never lose your work.
Usage Analytics
Per-model token breakdown, cost tracking, daily heatmap, session-level stats. Know exactly what you spend across all providers.
Team Dashboard
Read-only view into Claude Code multi-agent teams — task lists, teammate status, message flow, and real-time collaboration.
Plugin Marketplace
Browse, install, and manage Claude Code plugins, skills, and MCP servers from a visual marketplace with categories and search.
And more
Rich features built for developers
Inline Permissions
Rich permission review UI with Allow/Deny buttons and "Always Allow" rules.
Agent Management
Visual editor to create, edit, and manage custom agent definitions.
MCP Management
Discover MCP servers, view per-server status, reconnect or toggle from a panel.
Activity Monitor
Real-time hook event stream, tool activity timeline, subagent tracking.
Providers
15+ Supported Providers
Use Claude Code with any Anthropic-compatible API endpoint. Hot-switch between providers without restarting.
Supports any Anthropic-compatible API endpoint. Add your own with a custom base URL and API key.
Quick Start
Get started in minutes
Download
Grab the latest release from GitHub — DMG for macOS, ZIP for Windows.
Setup Wizard
On first launch, the wizard detects Claude Code CLI and configures authentication.
Choose a Provider
Pick from 15+ pre-configured providers or add a custom endpoint with your API key.
Start Coding
Open a project and start AI-assisted vibe coding with visual feedback.
FAQ
Frequently asked questions
Everything you need to know about OpenCovibe.
Is OpenCovibe free?
Yes, OpenCovibe is open-source under the Apache License 2.0. Free to use, modify, and distribute.
Does it require a cloud account?
No. OpenCovibe is local-first — all data is stored on your machine at ~/.opencovibe/. You only need network access for the AI model API calls.
Which AI models are supported?
OpenCovibe wraps Claude Code CLI, which supports Claude models via Anthropic API. Through third-party providers, you can also use DeepSeek, Kimi, Zhipu, MiniMax, DouBao, and any Anthropic-compatible model.
Do I need Claude Code CLI installed?
Yes. OpenCovibe wraps the CLI — it provides the visual interface while Claude Code handles the AI coding. The setup wizard will help you install it if needed.
Which platforms are supported?
macOS (primary, fully tested), Windows (functional, actively improving), and Linux (functional, community contributions welcome).
Can I use it without an Anthropic API key?
Yes! You can use any of the 15+ supported third-party providers. Each has its own API key. You can also use Ollama for fully local, offline AI coding.
Ready to upgrade your
coding workflow?
Download OpenCovibe and start AI-assisted vibe coding with a visual interface today.