Agent Zero BMAD K8s
Deploy a full AI agile development team on Kubernetes in minutes.
Live demo available β Try the GMeet interface to see 5 agents collaborating in real time.
What is BMAD K8s?
Agent Zero BMAD K8s deploys 9 specialized AI agents as a collaborative agile development team, running on Kubernetes (K3s) or Docker Compose. Each agent has a distinct persona, role, and tuned parameters β from a creative Analyst to a precise QA Engineer.
The agents communicate through Agent Zero's A2A protocol and are orchestrated by a central BMAD Orchestrator that tracks project phases, routes tasks, and enforces quality gates.
Key Features
- 9 Specialized Agents β Orchestrator, Analyst, PM, Architect, UX, Product Owner, Scrum Master, Developer, QA
- BMAD Method β Structured 4-phase workflow: Analysis β Planning β Solutioning β Implementation
- GMeet Dashboard (6 Tabs) β Meet, Dashboard, Config, Projects, Diary, Deploy β a full control plane for your agent team
- π§ Brain Viewer β Real-time SSE streaming of agent thinking: tool calls, reasoning, code execution, subagent delegation β with snapshot & share
- Markdown Transcript β Agent responses render with full markdown: code blocks, tables, lists, headings
- Live Monitoring β Agent health, response times, container CPU/memory stats, auto-refreshing dashboard
- In-Browser Config β Edit API keys, model settings, per-agent temperature without SSH
- Container Management β Start, stop, restart Docker containers and view logs from the Deploy tab
- Agent Diary β Shared activity feed with ping system, status snapshots, and filterable logs
- One-Line Installer β Deploy on any Linux server with a single
curlcommand - Kubernetes Native β Full K8s manifests with ConfigMaps, Secrets, PVCs, and Ingress
- OpenRouter Integration β Use any model (Kimi K2.5, Gemini Flash, Claude, GPT-4o, etc.)
- CI/CD β GitHub Actions for linting, validation, and build checks
Architecture Overview
ββββββββββββββββββββββββββββββββββββββββββββββββββββ
β GMeet Frontend β
β (React + Tailwind + Zustand) β
ββββββββββββββββ¬βββββββββββββββββββββ¬ββββββββββββββββ
β REST API β WebSocket
ββββββββββββββββΌβββββββββββββββββββββΌββββββββββββββββ
β GMeet Backend β
β (Express + WS on ports 4000/4001) β
ββββ¬ββββββββ¬ββββββββ¬ββββββββ¬ββββββββ¬βββββββββββββββββ
β β β β β X-API-KEY auth
ββββΌβββ ββββΌβββ ββββΌβββ ββββΌβββ ββββΌβββ
βBMAD β βMary β βJohn β βFred β βDana β ... (+4 more)
β:5001β β:5002β β:5003β β:5004β β:5005β
βββββββ βββββββ βββββββ βββββββ βββββββ
Agent Zero instances (Docker / K8s pods)
Quick Start
One-Line Install (K3s)
curl -sfL https://k8.virtualgpt.cloud/install.sh | bash
This installs K3s, clones the repo, prompts for your OpenRouter API key, creates Kubernetes secrets, and deploys all 9 agents.
Docker Compose (Local Dev)
git clone https://github.com/t4tarzan/agent0-bmad-k8.git
cd agent0-bmad-k8
cp .env.example .env.live # add your API keys
docker compose -f docker-compose.live.yml up -d
Next Steps
- Agent Profiles β Learn about each agent's role and capabilities
- BMAD Method β Understand the 4-phase workflow
- Deployment Guide β Detailed K3s and Kubernetes setup
- GMeet Dashboard β 6-tab control plane: Meet, Dashboard, Config, Projects, Diary, Deploy
- π§ Brain Viewer β Real-time agent thinking with SSE log streaming, snapshots & sharing
- API Reference β All backend endpoints including Brain, Diary, Deploy, and WebSocket