DevOps AI Toolkit

# Agents Architecture ## System Diagram ```mermaid graph TD Person[Person] ClientAI[Client AI Agent] ArtifactHub[ArtifactHub] subgraph "DOT AI" DotAI[Agent] Capabilities[Capabilities] Patterns[Patterns] Policies[Policies] VectorDB[(Vector DB)] MCP[MCP] REST[REST API] end subgraph Kubernetes K8sAPI[Kubernetes API] Kyverno[Kyverno] Controller[Controller] end Person -->|intent| ClientAI ClientAI -->|response| Person ClientAI -->|intent| MCP MCP -->|intent| DotAI DotAI -->|response| MCP MCP -->|response| ClientAI REST -->|intent| DotAI DotAI -->|response| REST DotAI -->|semantic search| VectorDB DotAI -->|operations| K8sAPI DotAI -->|discover charts| ArtifactHub K8sAPI -->|event| Controller Controller -->|intent| REST Capabilities -->|embeddings| VectorDB Patterns -->|embeddings| VectorDB Policies -->|embeddings| VectorDB Policies -->|enforce| Kyverno ``` ## Description This diagram illustrates the relationships between agents in the system. ### Components - **Person**: End user interacting through natural language - **Client AI Agent**: MCP-enabled coding assistant (Claude Code, Cursor, VS Code) - **MCP**: Model Context Protocol interface for AI tool integration - **REST API**: HTTP endpoints for CI/CD pipelines, controllers, and automation - **Agent**: Core AI-powered orchestration engine - **Capabilities**: Semantic descriptions of cluster resources stored in Vector DB - **Patterns**: Organizational deployment templates with best practices - **Policies**: Governance rules that guide recommendations and optionally enforce via Kyverno - **Vector DB**: Qdrant database for semantic search across capabilities, patterns, and policies - **Kubernetes API**: Cluster operations (deploy, remediate, operate) - **Kyverno**: Policy enforcement engine for active governance - **Controller**: [dot-ai-controller](https://github.com/vfarcic/dot-ai-controller) for event-driven remediation - **ArtifactHub**: External Helm chart repository for third-party tool discovery