DOCSEARCHbyAlgolia
Submit Docs

dot-ai DevOps AI Toolkit

repository·main·Indexed 43 days ago

https://github.com/vfarcic/dot-ai

An AI-powered development productivity platform and DevOps toolkit for platform engineering and Kubernetes operations. It enables natural language interaction with cloud-native infrastructure, providing capabilities for cluster operations, AI-powered issue remediation, semantic search over organizational documentation, and governance. The platform includes a REST API for tool execution and prompt management, and supports the Model Context Protocol (MCP) for integration with AI coding assistants.

Tokens
252.7K
Snippets
563
Records
1.1K
Agent score
56%

What's inside dot-ai

  1. Overview of DevOps AI Toolkit

    main

    The DevOps AI Toolkit is an AI-powered platform engineering and DevOps automation platform. It provides intelligent Kubernetes operations and conversational workflows to assist with cluster management and development.

    Key Capabilities:

    • Cluster Operations: Natural language querying, exploration, and intelligent Kubernetes deployment recommendations.
    • Remediation: AI-powered issue remediation, root cause analysis, and an Untrusted-content boundary to separate cluster output from operator instructions.
    • Knowledge Management: Semantic search over organizational documentation and shared prompt libraries.
    • Governance: Organizational pattern and policy management, and automated repository setup with governance files.

    Access Methods:

    • MCP (Model Context Protocol): Use this to integrate the toolkit with AI coding assistants.
    • CLI: Use this for direct agent integration.

  2. Overview of the Operate feature

    main

    The operate feature provides AI-powered Day 2 operations for Kubernetes resources using natural language intents. It allows you to update, scale, enhance, and manage workloads, databases, infrastructure, and cloud resources (via operators) with cluster-aware recommendations.

    Use this feature when you need to:

    • Perform operational changes on deployed resources (applications, databases, storage).
    • Manage cloud resources (AWS/Azure/GCP) via Kubernetes operators.
    • Manage networking or any Kubernetes-managed infrastructure.

    Available Interfaces:

    • MCP (Model Context Protocol)
    • CLI
    • Web Dashboard
    • Headlamp Plugin

  3. Overview of the Remediate tool capabilities

    main

    The Remediate tool provides AI-powered Kubernetes issue analysis and resolution. Key features include:

    • Root Cause Analysis: A multi-iteration investigation loop to understand complex Kubernetes issues.
    • Helm-aware Diagnostics: Detects issues with Helm-managed applications, such as stuck releases or failed upgrades.
    • GitOps-aware Remediation: For resources managed by Argo CD or Flux, the tool clones the source repository and creates GitHub pull requests with corrected manifests instead of using kubectl commands.
    • Dual Execution Modes: Supports a Manual Mode (human approval workflow) and an Automatic Mode (execution based on confidence/risk thresholds).
    • MCP Integration: Can augment investigations using tools from external Model Context Protocol (MCP) servers (e.g., Prometheus metrics or distributed traces).
    • Security: Features an always-on Untrusted-content boundary that wraps all tool results in explicit delimiters to prevent prompt injection/instruction following from tool data.

  4. What is instrumented in dot-ai?

    main

    The following components are automatically instrumented with OpenTelemetry, using official semantic conventions where applicable:

    • MCP Tools: All core tools (recommend, version, manageOrgData, remediate, projectSetup) are wrapped in INTERNAL spans.
    • AI Providers: Chat, tool loops, and embeddings operations use official gen_ai.* semantic conventions (e.g., gen_ai.tool.name, gen_ai.request.model, gen_ai.embeddings.count).
    • Kubernetes: Both the Kubernetes client library (via a JavaScript Proxy) and kubectl CLI executions are traced.
    • Vector Database: All Qdrant operations (e.g., search, upsert, deleteDocument) are instrumented with db.system: 'qdrant' attributes.
    • HTTP/REST: Entry point requests are captured as SERVER spans with W3C Trace Context support for distributed tracing.

  5. Overview of the DevOps AI Toolkit Project Website

    main

    The DevOps AI Toolkit is implementing a dedicated project website and documentation portal to improve feature discovery and user onboarding. The website is designed as a single-page site with an integrated documentation portal that leverages existing markdown files directly from the repository.

    Key features of the planned portal include:

    • Landing Page: Showcases project value propositions and key features (Cluster Discovery, AI Recommendations, Operator Integration, and Remediation).
    • Documentation Portal: Provides structured access to Getting Started guides, User Guides, API References, Examples, Architecture, and Contributing instructions.
    • Search Functionality: Enables users to quickly find specific technical information.
    • Repository-Centric: All documentation source files remain in the main repository to ensure version control and ease of maintenance.

  6. Overview of the CRD-Based Solution Tracking System

    main

    The CRD-based solution tracking system moves beyond session-based storage to provide persistent, Kubernetes-native solution management.

    Key Capabilities:

    • Persistence: Solutions and their states persist across sessions.
    • GitOps Integration: Designed for compatibility with GitOps tools like ArgoCD and Flux.
    • Lifecycle Management: Tracks solution states (e.g., active, deployed, failed, deleted).
    • Cluster Native: Uses standard Kubernetes patterns and RBAC for secure access.

    Planned Roadmap:

    • Phase 1: Core CRD implementation, schema design, and controller development.
    • Phase 2: GitOps integration, cross-cluster synchronization, and solution versioning.
    • Phase 3: Enterprise features including backup/disaster recovery, audit logging, and monitoring.

  7. Helm Tooling Capabilities and Constraints

    main

    The DevOps AI Toolkit provides Helm Day-2 operations through its operate and remediate tools. These tools allow an AI agent to manage the lifecycle of Helm releases.

    Supported Operations:

    • Upgrades: Handled via helm_install. It uses helm upgrade --install with the --reuse-values flag enabled by default to ensure safe upgrades.
    • Rollbacks: Handled via helm_rollback with revision selection.
    • Uninstalls: Handled via helm_uninstall.
    • Investigation: Includes helm_list, helm_status, helm_history, and helm_get_values to query release state.
    • Validation: helm_install_dryrun is available for safe analysis-phase validation before applying changes.

    Constraints and Out-of-Scope:

    • Repositories: Only public Helm repositories are supported (no private repos).
    • Secrets: Does not support encrypted values files (e.g., helm-secrets plugin).
    • Scope: Does not support multi-release operations or Helm chart development.
    • Registry: Initially supports only HTTP/HTTPS Helm repositories (no OCI registry support).

  8. Roadmap for Intelligent CI/CD Workflow Generation

    main

    The Intelligent CI/CD Workflow Generation feature is planned to evolve through several phases:

    Phase 2: Additional CI Platforms

    Support will be added for:

    • GitLab CI
    • Jenkins Pipeline
    • CircleCI
    • Azure DevOps Pipelines

    Phase 3: Advanced Features

    • Multi-environment deployments (dev, staging, prod)
    • Deployment approvals and gates
    • Rollback workflows
    • Notification integrations (Slack, Teams)
    • Cost optimization recommendations

    Phase 4: GitOps Integration

    • Auto-updating image tags in Helm values
    • Auto-updating Kustomize overlays
    • Triggering Argo CD/Flux sync after push

  9. Deploy infrastructure with dot-ai

    main

    Beyond application deployment, dot-ai supports deploying core infrastructure components including databases, networking, storage, and monitoring. The system uses an intelligent recommendation engine to map user intents (e.g., "PostgreSQL with high availability") to specific Kubernetes resources and cloud-provider patterns.

    Supported Infrastructure Categories

    • Databases: PostgreSQL (including HA configurations), Redis cache, and MongoDB operators.
    • Networking: Ingress controller setup, Load balancer configuration, and Network policy implementation.
    • Monitoring: Prometheus operator deployment, Grafana dashboard setup, and AlertManager configuration.
    • Operator Management: Installation and configuration of various Kubernetes operators.

  10. Review the Documentation Testing Scope

    main

    The current version of the documentation testing tool includes the following capabilities:

    In Scope

    • Single document validation: Processing documents section-by-section.
    • MCP interface: A complete feature set via the Model Context Protocol.
    • Session management: Ability to manage sessions and resume workflows.
    • Two-phase validation: Combined functional and semantic testing.
    • Structured results: JSON output format for programmatic analysis.
    • User-driven fixes: Selection and application of recommendations (Note: implementation status may vary by version).

    Out of Scope (Future Versions)

    • Real-time collaboration between multiple users.
    • Visual documentation testing (screenshot/UI validation).
    • Direct integration with platforms like GitBook or Confluence.
    • Advanced analytics dashboards for quality metrics.

  11. Capabilities of the Agentic Documentation Validation & Remediation System

    main

    The Agentic Documentation Validation & Remediation System is designed to automate the discovery, validation, and fixing of documentation. It operates within a dedicated Pod environment to ensure all necessary runtimes are available for code execution and syntax checking.

    Core Capabilities

    • Documentation Discovery: Supports cloning Git repositories to find markdown files or crawling URL-based documentation sites via sitemaps and link following.
    • Page Selection: Allows users to select specific pages, ranges of pages, or all pages for analysis.
    • Multi-Language Validation:
      • Syntax Validation: Supports JavaScript, TypeScript, Python, Go, YAML, JSON, and Bash.
      • Code Execution: Extracts code blocks and executes them to detect runtime errors.
      • Readability Analysis: Uses AI to analyze text readability and suggest specific improvements.
    • Automated Remediation:
      • Applies direct syntax fixes.
      • Generates AI-powered readability fixes.
      • Automatically creates Git branches, commits changes, and opens Pull Requests (PRs) with descriptive summaries.
    • Feedback Loop: Ingests reviewer feedback via MCP, CLI, or REST to apply corrections, revert selective fixes, and prevent repeating previously rejected approaches.
    • Session Management: Stores all sessions as JSON following dot-ai patterns, allowing sessions to persist beyond the lifetime of the execution Pod.

  12. Overview of Application Lifecycle Management in dot-ai

    main
    The dot-ai Application Lifecycle Management (ALM) system handles post-deployment operations for applications. This includes managing updates, scaling, configuration changes, troubleshooting, and deletion. The system is designed to provide zero-downtime updates and automated scaling to reduce manual operational tasks.