VibeiDE

About VibeiDE

VibeiDE is a desktop application for queueing and reviewing tasks from multiple AI coding agents in parallel. It runs Claude Code, Codex, and OpenCode in separate embedded terminals using your own CLI installations and provider accounts. The tool focuses on the review step of agent-driven development, keeping finished tasks in a pending state until a human marks them complete.

Review

VibeiDE addresses a specific friction point that emerges when a developer juggles several coding agents across different projects. Rather than adding another layer of agent orchestration, it gives each task its own session and terminal, queues work per provider, and holds results for manual sign-off. The maker built it for personal use and logged 386 hours in the tool over a 30-day period before launch.

Key Features

• Project queues with provider slots: You set a maximum number of parallel tasks per provider. A queued task starts automatically when that provider has a free slot.
• Per-task sessions: Every task runs in a fresh CLI session. Finished tasks wait for review, and you can send follow-up instructions within the same session instead of starting over.
• Plan council: One agent drafts a plan, and a second agent reviews it before any code is written. The maker notes this catches mistakes neither agent would catch alone.
• Git worktree support: Tasks can run on a single branch or each get their own worktree, which keeps parallel changes from tangling together.
• Dual view modes: You can watch a task in its original CLI terminal or switch to a chat view of the same session. A grid layout shows multiple terminals at once, and a 3D "Office" view arranges agents as desks.

Pricing and Value

VibeiDE runs on macOS and Windows, with Linux listed as experimental. A free 14-day trial is available without requiring a credit card. Launch pricing is set at $7.99 per month, $79 per year, or a $249 lifetime option. These launch prices are stated to last until January 1. The tool uses your own CLI installations and provider accounts and does not route around provider rate limits. Model costs are separate-OpenRouter's free models with tool calling let users run OpenCode tasks at no model cost, but Claude Code and Codex usage incurs charges directly through your own API keys.

Pros

• Finished tasks wait for human review by default, which reduces the risk of unverified changes landing in a codebase-especially relevant for auth, migrations, or database policy work.
• The per-task session model means you don't lose context; follow-ups continue exactly where a task stopped.
• Works with your existing CLI tools and provider accounts rather than wrapping them in a new abstraction layer.
• Plan council runs a second agent as a reviewer before code generation, catching planning errors early.
• Git worktree support isolates parallel tasks into separate working directories when needed.

Cons

• Diff review is currently limited to basic git preview; a more advanced per-task diff viewer is on the roadmap but not yet available.
• The tool is not well suited for developers who prefer a single-agent workflow and don't need parallel task management or a structured review queue.
• Linux support is experimental, so teams running Linux as their primary development OS may encounter stability issues.

VibeiDE fits solo developers and small teams who already use CLI-based coding agents and find themselves losing track of which task finished where. The review-first design adds a clear checkpoint that's absent from raw terminal workflows. It's less useful for developers who work with a single agent at a time and don't feel the tracking overhead that comes with running several in parallel.