DevOps AI Toolkit MCP Server for Developer Productivity

Overview

The DevOps AI Toolkit MCP Server implements the Model Context Protocol (MCP) to integrate AI-driven assistants into developer and DevOps workflows. It acts as a local, extensible service that exposes a set of “tools” — adapters to source code, CI/CD, Kubernetes, logs, and more — and mediates requests between an LLM or other model and the tools that provide domain data and actions. The goal is to enable safe, repeatable automation and contextualized AI assistance without embedding sensitive infrastructure access directly in models.

For teams, an MCP server simplifies building AI assistants that can answer codebase questions, generate deployment manifests, triage incidents, or operate CI pipelines by supplying scoped context and controlled tool access. It reduces boilerplate integration work and centralizes authentication, auditing and tool definitions so models can be reused across projects while keeping operations auditable and secure.

Features

• Exposes MCP-compatible endpoints for model-driven agents and local integrations
• Pluggable tools for source code, Git, CI/CD systems, Kubernetes, logs, and monitoring
• Configurable authentication and API-key support for model providers (OpenAI, local LLMs)
• Runs as a standalone binary, Docker container, or in Kubernetes
• Simple REST API for tool invocation, context inspection and result retrieval
• Audit-friendly — centralized point for logging actions and tool calls

Installation / Configuration

Quick start with Docker:

docker run -d \
  --name mcp-server \
  -p 8080:8080 \
  -e MCP_PROVIDER=openai \
  -e OPENAI_API_KEY="${OPENAI_API_KEY}" \
  -e MCP_AUTH_TOKEN="replace-with-secret" \
  vfarcic/dot-ai:latest

Run locally from source (Go):

git clone https://github.com/vfarcic/dot-ai.git
cd dot-ai
go build ./cmd/mcp-server
./mcp-server --port 8080 --auth-token replace-with-secret

Kubernetes manifest (basic):

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mcp-server
spec:
  replicas: 1
  selector: { matchLabels: { app: mcp-server } }
  template:
    metadata: { labels: { app: mcp-server } }
    spec:
      containers:
        - name: mcp-server
          image: vfarcic/dot-ai:latest
          env:
            - name: MCP_PROVIDER
              value: "openai"
            - name: OPENAI_API_KEY
              valueFrom:
                secretKeyRef:
                  name: openai-secret
                  key: api-key
            - name: MCP_AUTH_TOKEN
              value: "replace-with-secret"
          ports:
            - containerPort: 8080

Common environment variables

Available Tools / Resources

The server ships with or supports adapters for common DevOps resources:

• Git / repository search — file content and repo-wide queries
• CI/CD — trigger builds, get pipeline status
• Kubernetes — list resources, apply manifests, fetch logs (kubectl adapter)
• Logs & monitoring — query centralized logs or metrics
• Shell / command execution — run controlled commands in sandboxed environments
• Custom webhooks — call external services (issue trackers, chatops)

Each tool exposes a well-defined minimal API (name, description, inputs, outputs) that models can call through MCP messages. You can extend the server by adding custom adapters that follow the same interface.

Use Cases

1. PR summaries and change impact:

   • A model requests the repo tool to fetch changed files in a branch, then generates a concise PR description and a checklist of affected services.

2. Deployment manifest authoring:

   • An assistant asks for service metadata (image tags, env vars) from CI and secrets vault, composes a Kubernetes Deployment and a Helm values snippet, and returns it for review.

3. Incident triage:

   • On an alert, a runbook agent queries logs and Kubernetes pod status, compiles top suspect causes, suggests immediate mitigations (restart pod, scale replica), and drafts a timeline for Slack.

4. Automated housekeeping:

   • Periodic jobs call the server to identify stale feature branches, auto-generate cleanup PRs, or refresh dependency pinning using the Git and CI tools.

Example: invoking a tool via REST (replace token and URL):

curl -X POST "http://localhost:8080/v1/invoke" \
  -H "Authorization: Bearer replace-with-secret" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "git.search",
    "params": { "query": "TODO", "path": "src/" }
  }'

Security notes: protect MCP_AUTH_TOKEN, restrict network access to the server, and scope tool permissions to the minimum required for the assistant’s tasks.

Repository and issues: https://github.com/vfarcic/dot-ai

Tags: developer-tools