SkillFoxx

DevOps AI Toolkit MCP

The DevOps AI Toolkit MCP drives Kubernetes operations: cluster queries, deployment recommendations and root cause analysis.

  • 340 ★ on the repository
  • Last commit
  • MIT
MCP server

Broad access

Why this level

  • The agent deploys applications and remediates issues in a Kubernetes cluster.
  • Cluster output, including configurations and logs, goes into the model context.
All reasons and checks

vfarcic/dot-aiAuthor:vfarcic

Install

Manual install

Text for your agent

Deploy the engine following the guide and connect its address over HTTP in `.mcp.json` with a token or OAuth. Verify with the request "Show dot-ai status" and confirm cluster changes with the user.

Access and checks

How we review

Broad accessThe tool can change data, run commands or handle secrets.

  • The agent deploys applications and remediates issues in a Kubernetes cluster.
  • Cluster output, including configurations and logs, goes into the model context.

This is third-party code. Review the repository files before installing.

What it does

The DevOps AI Toolkit (dot-ai) brings platform engineering and Kubernetes knowledge to coding assistants. The engine is deployed into a Kubernetes cluster and clients connect to it over MCP from Claude Code, Cursor, VS Code, Cline and others, or through the CLI. Capabilities: natural language cluster queries and exploration, application deployment recommendations, issue remediation and root cause analysis, management of organizational patterns and policies, semantic search over documentation, repository setup with governance files and shared prompt libraries. An untrusted-content boundary separates cluster output from operator instructions. Connection is over HTTP with the `DOT_AI_AUTH_TOKEN` token or OAuth, which is on by default.

Who it is for. Platform engineers and Kubernetes administrators who want to run deployments and incident analysis from Claude Code, Cursor or VS Code.

Good fit when

  • The agent browses cluster resources in natural language and finds a faulty pod.
  • Deployment recommendations for a new app are requested with organization policies applied.
  • After a failure the agent works out the cause and proposes a fix.
  • A new repository is set up from a template with governance files and shared prompts.

Not a fit when

  • There is no Kubernetes cluster for the engine: the server deploys into one.
  • The agent must not change the cluster: deployment and remediation advice leads to real operations.
  • A lightweight client without a cluster deployment is needed: this engine requires the documented stack.

Example request

Show the cluster state, find pods in error, work out the cause and propose a fix, but do not change anything yet.

MCP

Transport
http
Authentication
OAuth

Environment variables

DOT_AI_AUTH_TOKEN secret
Static access token for the engine; OAuth can be used instead

Limitations

The engine is deployed into a Kubernetes cluster first and clients connect to its address, for example `http://dot-ai.127.0.0.1.nip.io` in the local stack example. For production with TLS use `https://`. Sign-in uses the static `DOT_AI_AUTH_TOKEN` token or OAuth: Cline does not support OAuth and Cursor supports it partially. The project collects anonymous usage statistics and opting out is described in the documentation. Remediation and deployment change cluster state, so confirm such steps.

How to disable. Remove the `dot-ai` entry from `.mcp.json` or the client settings and rotate the `DOT_AI_AUTH_TOKEN` token in the deployment. The engine stays in the cluster and is removed following the deployment guide.

FAQ

How do I connect a client?

Create `.mcp.json` with a `dot-ai` server of type `http`, the engine address and an `Authorization: Bearer <token>` header. OAuth needs no header.

Where do I get the token?

The `DOT_AI_AUTH_TOKEN` token comes from the deployment with `echo $DOT_AI_AUTH_TOKEN`.

How do I verify the connection?

Ask the client "Show dot-ai status": the answer shows the Kubernetes connection, the vector database and the tool list.

Which clients support OAuth?

Claude Code, ChatGPT, Windsurf and VS Code Copilot 1.101 and newer. Cursor supports it partially and Cline does not.

Is there an alternative to MCP?

Yes, a command line with lower token overhead for agents, scripts and CI.

Discussion

No questions or reviews yet.

MCP servers
Official

An AI assistant for Kubernetes that turns plain-language requests into kubectl operations and doubles as an MCP server

CLIBroad access7.6K

A Kubernetes panel with an MCP server: Claude Code reads pod logs, edits resources and rolls back releases in connected clusters.

MCP serverBroad access116

MCP server for Kubernetes: the agent reads cluster state, diagnoses pod failures and runs operations through kubectl

MCP serverBroad access972

A community MCP server for Azure DevOps work items, code and builds, with PAT authentication for on-premises Azure DevOps Server

MCP serverBroad access400

CLI and app for workspaces: each AI agent gets its own container from a shared devcontainer.json, with an MCP server and a skill

CLIBroad access137

README in short

The README lists capabilities, points to the stack and deployment guides and to the MCP setup page and mentions anonymous usage statistics.

README badge

Report a problem