The capability family
Six ways one app
changes how you ship
The Vibe Father isn't a chatbot with a terminal bolted on. It's a command deck — a team of agents, a brain that runs on any model, a gate that keeps them honest, and the plumbing to connect it all to your real stack.
A dev team, not a chatbot
Coordinator, Builder, Scout and Reviewer — each its own CLI and model — collaborating through a shared task board, with git worktrees so parallel work never collides.
- 4 specialist roles, mixed models
- Shared task board you can watch
- Parallel-safe via git worktrees
Any model. One mind.
Point Vibe at Anthropic, OpenAI, xAI, Kimi, MiniMax, OpenRouter or any OpenAI-compatible endpoint. Keys live in your Keychain; memory persists across every project.
- 11 provider presets + custom
- Borrowed-brain mode needs no key
- Persistent cross-project memory
Talk to your team
Dictate tasks and hear results read back. Speech recognition runs on-device with Apple's Speech framework — your voice doesn't need to leave your Mac.
- On-device speech recognition
- Speaks with macOS system voices
- Optional cloud TTS, your key
Connect once, just ask
Nineteen services — GitHub, Stripe, Supabase, Slack, your Postgres — plus any MCP server. Secrets stay in the Keychain and risky actions ask before they run.
- 19 services + custom MCP
- Secrets in the macOS Keychain
- Risky actions gated, logged
No AI grades its own homework
The AutoVibe gate detects your stack and runs your real build and tests before work reaches you. Checkpoints make every change rewindable — even the rewind.
- Real tests must pass, any CLI
- One-click checkpoints & rewind
- Autonomy modes in plain English
Let the best model win
Live cross-model benchmarks, a leaderboard and head-to-head compares — so you assign each role the model that actually earns it, not the one with the loudest launch.
- Live coding leaderboard
- Head-to-head model compares
- Published methodology
Also inside
The rest of the deck
The flagships get their own pages. These ship in the same app, at the same price.
Terminal grid
Run ~15 detected coding CLIs side by side in one window — Claude Code, Codex, Gemini, Kimi, OpenCode and more.
Click-to-edit preview
A live preview of your running app: point at any element and describe the change for an agent to make.
Automations
Recurring jobs your agents run on a schedule, without you babysitting a terminal.
SSH servers
Import your ~/.ssh/config read-only and let agents work on remote hosts — with a production confirmation gate.
Focus Mode
A distraction-free player for long agent runs: watch progress, skip the noise.
Vibe himself
A senior AI teammate that reads code, runs commands, browses, and coordinates the whole grid.
Every capability. One flat price.
All of this ships in one app — no tiers, no token meters, no per-seat tax. Bring the model subscriptions you already have.