# ViBeCoDeR (vbcdr) — Full Documentation > Terminal, browser, and devtools in one window per project — built for parallel AI coding. ## What is ViBeCoDeR? ViBeCoDeR (also written as vbcdr) is a free, open-source Electron desktop application that combines your terminal, browser preview, and developer tools into a single unified window per project. It eliminates constant context switching between terminal tabs, browser windows, and editor instances when working on multiple projects simultaneously with AI coding agents. It runs on macOS (Apple Silicon), Windows, and Linux. The full source is on GitHub under the MIT license. ## The problem it solves AI coding agents — Claude Code, Codex CLI, Aider, GitHub Copilot — let developers run several projects in parallel. But the surrounding tools weren't built for that workflow: - Terminals, browsers, and editors are separate apps. Finding the right tab for the right project means constant alt-tabbing. - localhost shares cookies across all ports. Three projects on three different ports break each other's auth redirects and session state. - Your mental model says everything belongs to one project, but in reality the context is scattered across multiple apps and windows. ViBeCoDeR collapses all of that into a single window per project, with isolation guarantees that make parallel development actually work. ## How it works ### 1. Open a project Drop any repo into ViBeCoDeR. It gets its own isolated workspace — terminals, browser context, layout, credentials, and theme all scoped to that project. ### 2. Prompt your agent Describe what you want in natural language. Drag in mockups, screenshots, or reference files as context. Kick off a long-running task and switch to another project while it works. ### 3. See it live, ship it Changes appear in the built-in preview instantly. Errors forward to your agent with one click. No more copy-pasting stack traces like a middleman. ## Features ### Project isolation Each project gets its own browser context, cookies, and localStorage. No more cross-port collisions between projects running on different localhost ports. Auth flows, session state, and storage stay scoped to where they belong. ### Works with any AI agent Compatible with Claude Code, Codex CLI, Aider, GitHub Copilot, and any other terminal-based AI coding agent. Run multiple agents per project if you want to compare outputs. ### Drag-and-drop context Drag files, screenshots, or mockups directly into terminals — they auto-attach as context for your agent. No more manual paste or save-then-reference dance. ### Multi-LLM terminals per project Run multiple LLM terminals within a single project. Each terminal is labeled by the Claude process currently running, so you always know what's happening where. ### Live terminal status indicators Per-terminal and per-project status indicators. A green dot appears when Claude finishes a task — context-switch to another project and check back when it's done. ### Inline diff overlay After each terminal command runs, an inline diff overlay surfaces exactly which files changed. No more switching to a side terminal to run `git diff` after every step. ### Task queue and command palette Queue prompts as inline chips, drag them to reorder, and let them auto-run when the agent goes idle. A ⌘K command palette jumps anywhere in the app and accepts prompt-to-queue input directly. ### Skills marketplace Browse, search, and install Claude skills with per-scope buttons. Pick a global or project-scoped install — no manual file shuffling. ### Commands editor Define custom commands and reusable prompts per project or globally. Project-scoped commands stay with the repo, global commands follow you everywhere. ### Network inspector See every request with headers, timing, and response sizes. Forward failures to your agent in one click instead of copy-pasting stack traces. ### Credential manager Per-project isolated credential stores for different environments and API keys (Convex, Clerk, custom). Switch projects and your credentials switch with you. ### Git visualization Commit graph with branch lanes and one-click commit actions. No more switching to GitHub to see what's going on with the branch you just pushed. ### Device emulation Preview your app in mobile, tablet, and desktop viewports with proper user agents. Catch responsive bugs without leaving the project window. ### Usage analytics A dedicated Usage page with a token-velocity sparkline, persisted history across sessions, and a range chart. The Sessions/Projects table is sortable across any timeframe so you can see what your agents actually cost. ### Statistics and dashboard A Statistics page with theme-aware charts and a Commits/Terminal/Both source toggle. The dashboard shows project cards with live terminal output buffers and a token-usage progress bar per project. ### Persistent terminals Terminal tabs and full scrollback survive app restarts. Pick up a long-running agent exactly where you left off, even after a reboot. ### Editor power-ups Drag-to-reorder editor tabs, tabs colored by git status, an editable diff editor, format-on-save, debounced autosave, configurable font and tab size, bracket pair colorization, and a minimap toggle. ### File tree operations Delete, rename, create, duplicate, and search files directly from the tree. Includes docx preview support. ### Themes 12 built-in themes — Dracula, Catppuccin, Nord, Tokyo Night, Afternoon, and more — each with dark and light variants. A live custom theme editor lets you build your own. ### Customization Configurable token cap, configurable background image with blur for app panels, live-toggle terminal transparency, and a selectable idle notification sound when your agent finishes. ## Multi-project workflow ViBeCoDeR is designed for the case where four or five projects are running in parallel, each with its own AI agent. Switching between them preserves terminal state, layout, scroll position, and theme. You can leave a long-running agent on Project A, jump to Project B for ten minutes, come back to a finished task with a green dot waiting. ## Technical details - Built with Electron - MIT licensed - macOS (Apple Silicon), Windows, and Linux - Source code: https://github.com/jestersimpps/vbcdr-electron - Latest release: https://github.com/jestersimpps/vbcdr-electron/releases/latest ## Frequently asked questions **Is ViBeCoDeR free?** Yes — fully free and open source under the MIT license. **Does it work with my AI agent?** If your agent runs in a terminal, yes. ViBeCoDeR has been tested with Claude Code, Codex CLI, Aider, and GitHub Copilot. **Will my localhost cookies still collide?** No. Each project has its own isolated browser context. Three projects on three ports each get their own cookies and localStorage. **Can I run multiple agents in one project?** Yes — multiple LLM terminals per project, each labeled by the running process. **Do my terminals survive a restart?** Yes — terminal tabs and full scrollback are persisted across app restarts. ## Links - Website: https://www.vbcdr.io - LLM-friendly summary: https://www.vbcdr.io/llms.txt - GitHub: https://github.com/jestersimpps/vbcdr-electron - Latest release (macOS / Windows / Linux): https://github.com/jestersimpps/vbcdr-electron/releases/latest - Author: Jo Vinkenroye — https://jovweb.dev - Sponsor: https://github.com/sponsors/jestersimpps - License: MIT