Skip to content

VibOpsai/vibops-install

Repository files navigation

VibOps — Installation

Deploy VibOps on your infrastructure in minutes. This repository contains everything needed to run VibOps — no source code, just configuration and pre-built Docker images.

Prerequisites

  • Linux server: 4 vCPU, 8 GB RAM, 50 GB SSD minimum
  • Docker 24+ and Docker Compose v2
  • An LLM API key (Claude recommended) or on-prem LLM (Ollama, vLLM)

No GPU required on the VibOps server — GPUs stay on your clusters.

Quick Start

# 1. Clone this repository
git clone https://github.com/VibOpsai/vibops-install.git
cd vibops-install

# 2. Authenticate to the VibOps container registry (token provided by VibOps)
make login VIBOPS_REGISTRY_TOKEN=<your-token>

# 3. Start the full stack (generates secrets, starts all services)
make quickstart

# 4. Set your LLM provider in .env
#    Edit LLM_PROVIDER and LLM_API_KEY, then:
docker compose restart agent

# 5. Create your organisation
make pilot-create-client ORG="My Company" EMAIL=admin@company.com PASSWORD=yourpassword

# 6. Open the console
#    http://SERVER_IP:8003

What's included

File Purpose
docker-compose.yml Full stack: core, agent, console, worker, PostgreSQL, Redis, Prometheus, Grafana
.env.example Environment template — all variables documented
Makefile Quickstart, health checks, client provisioning, backups
prometheus.yml Prometheus scrape configuration
alerting_rules.yml GPU and service alerting rules
grafana/ Grafana datasources and pre-built dashboards
scripts/ Health check, gateway setup, onboarding automation
docs/installation.md Complete installation guide (13 sections)

Services

Service Port Description
Console 8003 Web UI — open in browser
Core API 8000 REST API + job engine
Agent 8001 LLM agent
LLM Proxy 8004 OpenAI-compatible inference proxy — per-agent GPU cost attribution
Grafana 3000 Dashboards (admin / auto-generated password)
Prometheus 9090 Metrics

API Documentation (Swagger)

The full interactive API documentation is available at:

  • Swagger UI: http://SERVER_IP:8000/docs
  • ReDoc: http://SERVER_IP:8000/redoc

Both endpoints are enabled by default (APP_ENV=development). In production mode (APP_ENV=production), they are disabled and return 404 for security.

LLM Inference Proxy (per-agent FinOps)

VibOps includes a transparent OpenAI-compatible proxy that tracks GPU cost per AI agent. Point your agents at the proxy instead of your LLM server:

# Your agents call the proxy instead of vLLM/Ollama directly
OPENAI_BASE_URL=http://SERVER_IP:8004/v1

# Add a header to attribute costs per agent
curl -X POST http://SERVER_IP:8004/v1/chat/completions \
  -H "X-VibOps-Agent-Id: pricing-agent" \
  -H "X-VibOps-Team: supply-chain" \
  -d '{"model": "mistral:7b", "messages": [...]}'

Per-agent cost attribution is visible in the console under FinOps → Agent LLM Usage.

Connect a GPU cluster

GPU clusters connect to VibOps via an outbound-only gateway (no inbound ports required on the cluster side):

# From the VibOps console: Fleet → Add Gateway → copy the token
# On the GPU cluster:
helm upgrade --install vibops-connect vibops/vibops-connect \
  --set vibops.coreUrl="https://your-vibops-server" \
  --set vibops.token="<token-from-console>"

Licence

VibOps starts a 14-day trial automatically (10 GPUs / 5 users / 2 clusters). Contact david@vibops.ai for a production licence key.

Full documentation

See docs/installation.md for the complete guide including:

  • Helm production deployment
  • On-prem LLM (air-gapped / sovereign)
  • Team management and RBAC
  • Configuration reference
  • Upgrade and backup procedures

Troubleshooting

If something isn't working, generate a debug bundle and send it to VibOps support:

make debug
# → vibops-debug-2026-06-24-143000.tar.gz
# Send this file to david@vibops.ai

The bundle collects logs, container status, health checks, resource usage and configuration — all secrets are automatically redacted.

Support

About

Install VibOps — GPU infrastructure control plane with per-agent LLM cost tracking. Docker Compose, one command.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Packages

 
 
 

Contributors