DevOps AI Toolkit

# Quick Start **Get started with DevOps AI Toolkit in minutes - deploy applications, manage policies, and remediate issues using AI-powered Kubernetes workflows through MCP.** ## Overview **What it does**: DevOps AI Toolkit provides AI-powered Kubernetes deployment, remediation, policy management, and capability discovery through conversational workflows in your MCP-enabled coding agent. **Use when**: You want intelligent Kubernetes operations without memorizing commands, need AI-powered troubleshooting, or want to establish governance policies across your cluster. **📖 Full Guide**: See [MCP Setup Guide](mcp-setup.md) for detailed configuration options and [Tools Overview](mcp-tools-overview.md) for complete feature reference. ## Prerequisites **Works without AI keys:** - ✅ **Shared prompts library** - No API key needed, works with any MCP-enabled coding agent **For AI-powered features (deployment, remediation, patterns, policies, capabilities):** - **AI Model API key** - Required for AI analysis and intelligent recommendations - **Multiple AI models supported** - see [AI Model Configuration](mcp-setup.md#ai-model-configuration) for all options and setup - **Quick setup**: Claude (default) - `export ANTHROPIC_API_KEY=your_key_here` **For Kubernetes deployment recommendations:** - **kubectl** configured with cluster access - Verify cluster access with: `kubectl get nodes` - Should show your cluster nodes without authentication errors **For organizational pattern management:** - **Vector DB service** (Qdrant) for pattern storage and semantic search - **Embedding provider API key** - Required for semantic pattern matching: - OpenAI: `OPENAI_API_KEY` - Google: `GOOGLE_API_KEY` - Amazon Bedrock: AWS credentials via environment variables or `~/.aws/credentials` **For policy management and governance:** - **Vector DB service** (Qdrant) for policy storage and semantic search - **Embedding provider API key** - Required for semantic policy matching (same options as above) - **Optional**: Kyverno installed in cluster for active policy enforcement ## Installation DevOps AI Toolkit is designed to be used through AI development tools via MCP (Model Context Protocol). No direct installation needed - simply configure your AI tool to connect to the MCP server. ## Usage **🎯 Recommended: Docker Setup (Complete Stack)** Perfect for getting all features working immediately with minimal setup: 1. **Download Docker Compose configuration:** ```bash curl -o docker-compose-dot-ai.yaml https://raw.githubusercontent.com/vfarcic/dot-ai/main/docker-compose-dot-ai.yaml ``` 2. **Set environment variables and create MCP configuration:** ```bash # Set your AI model API key (example with Claude - see setup guide for other models) export ANTHROPIC_API_KEY="sk-ant-api03-your-key-here" # Set embedding provider key if needed (see setup guide for options) export OPENAI_API_KEY="sk-proj-your-openai-key-here" # Create MCP configuration for Claude Code cat > .mcp.json << 'EOF' { "mcpServers": { "dot-ai": { "command": "docker", "args": [ "compose", "-f", "docker-compose-dot-ai.yaml", "--env-file", ".env", "run", "--rm", "--remove-orphans", "dot-ai" ] } } } EOF ``` **What you get:** - ✅ **Complete Stack**: MCP server + Qdrant vector database included - ✅ **All Features Working**: Capability management, pattern storage, policy governance, semantic search - ✅ **No External Dependencies**: Everything runs in containers - ✅ **Kubernetes Integration**: Direct kubectl access to your clusters **Alternative Methods**: See the [MCP Setup Guide](mcp-setup.md) for NPX (Node.js) and Development setup options. 3. **Start your MCP client:** ```bash claude # or your preferred MCP-enabled AI tool # Verify everything works by asking: "Show dot-ai status" ``` 4. **Use conversational workflows:** **Example: Kubernetes Deployment** ```text User: I want to deploy a web application to my cluster Agent: I'll help you deploy a web application. Let me get recommendations. First, a few questions to provide better suggestions: What type of web application is this? Is it for development or production? Does it need external access? [Uses recommend tool for clarification] User: It's a production Node.js API that needs external access Agent: Perfect! Based on your production requirements, I found 3 deployment options. Let's use the Production Web Stack with external access. What's your application name and container image? [Uses recommend tool with refined intent] User: App name is "myapp" and image is "myapp:v1.0.0" Agent: Excellent! Generating production-ready manifests and deploying... [Uses recommend tool with stage routing: chooseSolution → answerQuestion → generateManifests → deployManifests] Agent: ✅ Successfully deployed! Your production API is running with external access. ``` **Example: Capability Management** ```text User: I want to scan my cluster for resource capabilities to improve recommendations [Uses manageOrgData with capabilities scanning] Agent: Would you like to scan all cluster resources or specify a subset? Options: 1) All resources 2) Specific subset User: All resources in auto mode Agent: Starting comprehensive cluster capability scan... ✅ Capability scan completed! Processed cluster resources successfully. User: Now I need a PostgreSQL database Agent: Let me get enhanced recommendations using your capability data. [Uses recommend tool with capability pre-filtering] Agent: Perfect! I found sqls.devopstoolkit.live as the top match - it's a managed database solution supporting PostgreSQL with multi-cloud capabilities and low complexity. Much better than the generic StatefulSet approach I would have suggested before! ``` **Example: Kubernetes Issue Remediation** ```text User: Something is wrong with my database in postgres-demo namespace Agent: I'll investigate the database issue using AI-powered analysis. [Uses remediate tool for systematic investigation] Agent: 🔍 Root Cause (95% confidence): PostgreSQL pod can't start because the required PersistentVolumeClaim 'postgres-pvc' is missing. 🔧 Recommended Fix: Create missing PVC - Risk Level: Low - Command: kubectl apply [PVC manifest] Choose execution: 1) Execute via MCP 2) Show commands User: 1 Agent: ✅ Issue Resolved! Created postgres-pvc, pod transitioned to Running state. Database is operational! ``` **Example: Policy Management** ```text User: I want to create a policy requiring resource limits on all deployments Agent: I'll help you create a governance policy for resource limits. Let me start the creation workflow. [Uses manageOrgData with policy creation] Agent: Please describe your resource limits policy intent: User: All container deployments must have CPU and memory limits to prevent resource exhaustion. CPU limits should be between 100m-2000m, memory between 128Mi-4Gi. Agent: Excellent! I've created your Resource Limits Policy and generated a Kyverno ClusterPolicy for enforcement. Would you like to: 1. Store policy intent only (for AI guidance) 2. Apply Kyverno policy to cluster (active enforcement) User: Apply the Kyverno policy for active enforcement Agent: ✅ Policy deployed! Now when users deploy applications, they'll get required questions for CPU/memory limits with your specified ranges, and any non-compliant deployments will be blocked by Kyverno. ``` *Note: Conversational examples are illustrative - actual AI responses will vary based on specific context and implementation.* **Example: Shared Prompts Library** ```text # Conversational approach User: I want to create a new PRD for a feature Agent: I'll help you create a documentation-first PRD. Let me start the process. [Uses prd-create prompt via /dot-ai:prd-create] Agent: Great! I've created GitHub issue #34 and the PRD file. What feature would you like to document? # Direct slash command approach User: /dot-ai:prd-create Agent: I'm executing the PRD creation workflow. Please describe the feature you want to create a PRD for... ``` ## Next Steps 📖 **[MCP Setup Guide →](mcp-setup.md)** - Detailed configuration, troubleshooting, and examples 📖 **[Complete Tools & Features Reference →](mcp-tools-overview.md)** - Comprehensive guide to all available tools, workflows, and advanced features