AI DevOps Platform

AI-powered DevOps control, natively inside VS Code.

Connect GitHub Copilot — or any MCP-compatible AI agent — directly to your Git, Docker, Kubernetes, CI/CD, Terraform, and Prometheus infrastructure. Ask questions and take action in plain English, without leaving your editor.

Why this exists

DevOps work today means switching between a dozen CLIs, dashboards, and browser tabs just to answer a simple question like "is my deployment healthy?" This extension collapses that entire workflow into a single conversation with your AI assistant.

"Is the latest deployment healthy, and did the pipeline pass?"
        │
        ▼
   AI reasons about which tools it needs
        │
        ▼
   Calls your Kubernetes + CI/CD tools automatically
        │
        ▼
   "Yes — all 3 replicas are available, and the last
    GitHub Actions run on main completed successfully."

No manual kubectl, no tab-switching to GitHub Actions, no context-switching. You ask, the AI investigates using real, live data, and reports back.

How it works

The extension runs a self-contained, Dockerized MCP (Model Context Protocol) gateway locally on your machine. It's fully automatic — install the extension, and the gateway is pulled, started, and health-checked with zero manual setup.

┌─────────────────────────────┐
│   GitHub Copilot Chat        │   ← you ask, in plain English
│   (Agent Mode)                │
└──────────────┬────────────────┘
               │  MCP (Model Context Protocol)
               ▼
┌─────────────────────────────┐
│   AI DevOps Platform          │   ← this extension
│   Extension (VS Code)         │
│                                │
│   • Activity Bar + Tree View  │
│   • Docker lifecycle manager  │
│   • MCP client                │
└──────────────┬────────────────┘
               │ manages
               ▼
┌─────────────────────────────┐
│   MCP Gateway (Docker)        │   ← auto pulled & started
│                                │
│   Auth • Rate Limiting •      │
│   Audit Logging • Confirmation│
│   Tokens for mutating actions │
└──────────────┬────────────────┘
               │
   ┌───────────┼────────────┬─────────────┬──────────────┐
   ▼           ▼            ▼             ▼              ▼
  Git       Docker      Kubernetes      CI/CD        Terraform
                                                            +
                                                       Prometheus

Getting Started

Install this extension from the VS Code Marketplace
Make sure Docker Desktop is installed and running
Open the AI DevOps Platform view from the Activity Bar — the gateway container is pulled and started automatically
Open GitHub Copilot Chat, switch to Agent mode
Start asking:
- "List my running Docker containers"
- "Are all pods healthy in the production namespace?"
- "Did the last CI run pass on main?"
- "Show me the current Terraform plan"

That's it — no manual server setup, no CLI configuration.

Supported Integrations

Domain	What you can ask
Git	Repository status, branches, commits, pull requests, PR checks
Docker	List/inspect containers & images, view logs, start/stop/restart/remove (confirmed before executing)
Kubernetes	Pods, deployments, services, nodes, namespaces, events, cluster health, scale/restart/delete (confirmed before executing)
CI/CD	Workflow status & history, job logs, releases, trigger/rerun/cancel runs (confirmed before executing)
Terraform	Workspaces, plan, state inspection, drift detection
Prometheus	Instant & range metric queries, active alerts, scrape target health

Security by Design

Every action that changes something — restarting a deployment, removing a container, triggering a pipeline — follows a strict two-step confirmation flow:

The AI previews the exact action and asks you to confirm, in plain language
Only after your explicit approval does the action actually execute

Nothing destructive ever runs from a single, possibly-ambiguous request. Read-only actions (status checks, listings, log viewing) run immediately with no friction.

Authentication is handled via OAuth 2.1 (Auth0), with per-tool scopes, per-subject rate limiting, and a full audit log of every tool call.

AI-DEVOPS-PLATFORM