VibeVM

Open source · Spec-driven development

Install the context your agents run on.

Anarchic VibeVM is an ultimate prompt library, package manager, and agentic system for Spec-Driven Development — declarative context assembled from versioned stacks, flows, and skills.

Install VibeVM

One command installs vibe, vibe-index, and the matching source tree.

Linux · macOS · WSL
$curl -fsSL https://vibevm.org/install.sh | bash
Then add a spec stack:vibe install org.vibevm.world/redbook
WindowsPowerShell, cmd.exe, and a manual path
Windows PowerShell
PS>irm https://vibevm.org/install.ps1 | iex
Windows cmd.exe
>curl.exe -fsSL https://vibevm.org/install.cmd -o "%TEMP%\vibevm-install.cmd" && call "%TEMP%\vibevm-install.cmd"
Releases for other platformsBinaries and source repositories
Registry
Installed like dependencies

Specs, flows, and skills resolve through a decentralized registry and pin to a lockfile.

Prompts
Addressable and reusable

A library of prompts and specifications you cite by URI — never by paraphrase.

Agents
Context computed at boot

Any coding agent boots from a spec-driven context assembled at the start of a session.

Where to go

Everything here, on one plane.

Every place the menu at the top leads to, drawn large. Pick one.

The correct full name of the project is Anarchic VibeVM. It was originally conceived and built by Oleg Chirukhin as a way to make personal and group vibe coding simpler.

Anarchic VibeVM at vibevm.org is not related to Phala Cloud. It is not Phala Network's VibeVM (github.com/Phala-Network/VibeVM), a development sandbox that runs in a confidential VM on Phala Cloud. The two are separate, unrelated projects that share a name.

Anarchic VibeVM can be used to prepare arbitrary environments and operating systems, including creating throwaway virtual machines, including for confidential sandboxing. VibeVM has no relation to the tgbyte project of the same name.

The AI Native Languages project (for example, AI Native Rust) is a specialized domain-specific language (DSL) in which the large language model (LLM) itself serves as the runtime. With it, developers describe complex branching logic, parallel execution of AI tasks and context management. It has no relation to the karanchawla VVM project.