Skip to main content

Vibecontrols Overview

Vibecontrols is the Agentic Engineering IDP for AI-Native teams.

It brings your machines, project contexts, terminal sessions, tunnels, and AI tools together in one place. Whether you are managing a handful of local projects or coordinating development across dozens of remote machines, Vibecontrols gives you a unified workspace for orchestrating where your work runs.

What Vibecontrols Does​

Modern development involves many moving parts: remote servers, multiple terminal sessions, project-specific environment variables, documentation, and AI assistants that need context about what you are working on. Vibecontrols organizes all of this around a small set of well-defined concepts.

You can think of it as your agentic engineering Internal Developer Platform — the place where your machines check in via an agent, your projects are first-class objects, sessions run safely shareable, and AI tools have grounded context for the code in front of them.

Core Concepts​

Targets​

A target is a connection destination — a machine where work can happen. Targets come in four flavors: DIRECT (the local machine), SSH (a remote host you can SSH into), RDP, and VNC. Each target stores host, port, username, and connection-type-specific config.

Agents​

An agent is the @vibecontrols/agent process running on a target. Once installed and started, it connects to the Vibecontrols backend over an outbound tunnel, exposes a local REST API for plugins and the CLI, and hosts sessions, tunnels, and AI workflows.

The agent reports heartbeat and capabilities. The backend tracks isActive, lastHeartbeat, version, hostname, platform, and the agent's own outbound tunnel URL.

Vibes​

A vibe is a named project context. Vibes are database records — there is no vibe.yaml to write — created via the CLI or GraphQL.

A vibe has a type (PROJECT, WORKSPACE, REPOSITORY, MONOREPO, PACKAGE, or CUSTOM), a path on the agent's filesystem, optional git metadata, environment variables, status (ACTIVE, INACTIVE, ARCHIVED, ERROR), and supports parent/child relationships for representing monorepos.

Sessions​

A session is a live terminal multiplexer or shell running on an agent. Vibecontrols supports seven session types — TMUX, WEZTERM, ZELLIJ, SSH, TERMINAL, SCRIPT, and CUSTOM — each implemented by a plugin.

Sessions are persistent across disconnects (for multiplexer types), get a browser-accessible URL via ttyd, and can be shared with teammates or via a token-protected link.

Notes​

Notes are Markdown content attached to vibes. They support tags, pinning, and search — designed to keep operational knowledge close to where the work happens.

Tunnels​

Tunnels expose ports on the agent's machine to the public internet without VPN or manual firewall configuration. Two providers are built in: Cloudflare quick tunnels (uses cloudflared, gives you *.trycloudflare.com URLs) and VibeTunnels (frpc-based, requires your own frps server).

VibeDecks​

A VibeDeck is a Stream-Deck-style grid of action buttons. Each button runs a command, script, or webhook on a target agent with one click. You configure the grid (columns, rows, button size, color theme) and assign each button a row/column position, an action type (COMMAND, SCRIPT, or WEBHOOK), and the actual payload.

VibeDecks are not bundles of vibes — they're action dashboards. Share a deck via a token-based share link (VIEW or EXECUTE permission).

Webhooks​

Webhooks notify your other tools when things happen — sessions start or stop, vibes are created, agents come online. Configure an endpoint + secret + event allowlist, and Vibecontrols delivers signed HMAC-SHA256 payloads with retry-on-failure.

AI Prompts and AI Sessions​

Vibecontrols has a built-in AI area where you can define reusable prompt templates, organize them by category (CODING, DEBUGGING, REVIEW, DOCUMENTATION, TESTING, DEPLOYMENT, RUNBOOK, etc.), substitute variables at render time, and dispatch them to AI providers running on your agents (Claude, OpenAI, OpenRouter). The AI plugin keeps prompt context tied to the machine where the work runs.

Configurations​

Configurations are scoped key-value settings. Scope can be WORKSPACE, VIBE, SESSION, AGENT, USER, or GLOBAL. Values marked isSecret: true are encrypted at rest.

Actions​

Actions are reusable command / script / webhook definitions that can run manually, on a schedule, or in response to an event. Each action execution captures output, exit code, duration, and status. VibeDeck buttons can reference actions for consistency across decks.

Who Vibecontrols Is For​

  • Individual developers who want a clean way to manage multiple project contexts across machines
  • AI-native teams who want their AI assistants grounded in real environments, not abstract code
  • DevOps and platform teams who want a Stream-Deck-style operations console for routine ops
  • Distributed engineering organizations who want to onboard, share, and audit work that spans many machines
  • Customers of Burdenoff Workspaces who want their dev environments to plug into the same identity, billing, RBAC, and audit fabric as the rest of their stack