Vibeleaderboard
Index / tool

pve-agents

pve-agents.sh
Visit pve-agents.sh
Category
AI Agents
Rank
No. 2282Tools index

Previous survey · No. 2132 ·

Pricing
Open Source
Type
TOOL
Use case
Agent Building · Coding
Interfaces
Web
GitHub
2 stars
Date

About

pve-agents takes a repository, a ref and a purpose, builds a disposable LXC container on your Proxmox host, and starts a Claude Code agent inside it. The agent transcript streams to the browser, where users answer permission prompts, review per-file diffs, and push to a branch or discard the work. Isolated containers keep agent actions off the user's workstation.

At a glance

pve-agents starts coding-agent jobs in disposable Proxmox containers and streams their work to a browser. It is worth evaluating if you already run Proxmox and want separate workspaces with a review step before pushing changes.

Good fit

Operators who can maintain a container template and restrict which repositories and credentials jobs can use.

Consider something else when

Pass if you need a managed runner or your isolation policy requires virtual machines for these workloads.

Start here

Launch a job against a public test repository and inspect the streamed log and diff. Discard the workspace and confirm cleanup affected only that container.

Cost and requirements

The AGPL-licensed controller needs a Proxmox host, template, scoped API access and credentials for the chosen agent. You provide compute and inference.

Limits to know

The browser interface can expose transcripts and an interactive shell. Review controller access, container network reachability and credential scope before running private work.

Tags

proxmoxlxcclaude-codesandboxself-hostedcoding-agent

Tech Stack

Node.jsTypeScriptVite

Media

pve-agents

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.