v1.23.1 - Code Optimization & Security Fixes

Changes

Security Fixes

• Fixed os.system() command injection in networking.py port-forward — replaced with subprocess.Popen using list args
• Fixed temp file leak in kubectl_apply — wrapped os.unlink in try/finally
• Fixed command.split() in kind_node_exec — replaced with shlex.split()
• Added non-destructive guard to exec_in_pod and port_forward
• Restricted kubectl_generic allowlist — config limited to safe subcommands, auth limited to can-i
• Fixed hardcoded version string in HTTP handler

DRY Consolidation

• Consolidated _get_kubectl_context_args from 10 duplicate implementations to single source in k8s_config.py
• Extracted shared Helm repo add/update helper
• Created _cli_utils.py with cached CLI availability checks and common subprocess runner
• Standardized non-destructive mode patterns across all tool files

Dead Code Removal

• Removed unused imports across 7 files
• Removed unused check_safety_mode decorator and is_operation_allowed function from safety.py
• Removed auto-pip-install antipattern from mcp_server.py

Code Quality

• Fixed bare except: clauses in networking.py and cost.py
• Fixed None values in pod recommendations list
• Proper exception chaining with from err in FastMCP import
• Updated test suite: 469 tests passing

Install

pip install kubectl-mcp-server==1.23.1
# or
npx kubectl-mcp-server@1.23.1

v1.23.0

Fix: In-Cluster Kubernetes Config

Resolves the HTTPConnectionPool(host='localhost', port=80): Max retries exceeded error when the server is deployed inside a Kubernetes cluster (#61).

What Changed

• Fixed config loading — _patched_load_kube_config no longer short-circuits calls with explicit arguments (config_file, context, client_configuration), which was preventing the kubernetes client from loading the correct API server address
• In-cluster fallback — The kubeconfig provider now automatically falls back to in-cluster config when no kubeconfig file exists in the container
• Actionable errors — Raises ProviderError with clear remediation steps when neither kubeconfig nor in-cluster config is available, instead of silently falling back to localhost:80
• Deployment manifest — Added MCP_K8S_PROVIDER=in-cluster and KUBERNETES_SERVICE_PORT=443 env vars

Deployment

For in-cluster deployments, ensure these env vars are set:

env:
  - name: MCP_K8S_PROVIDER
    value: "in-cluster"
  - name: KUBERNETES_SERVICE_HOST
    value: "kubernetes.default.svc"
  - name: KUBERNETES_SERVICE_PORT
    value: "443"

Reference manifests: deploy/kubernetes/

Install

pip install kubectl-mcp-server==1.23.0
# or
npx kubectl-mcp-server@1.23.0
# or
docker pull rohitghumare64/kubectl-mcp-server:1.23.0

v1.22.0 - kubectl-mcp-app Interactive UI Dashboards

What's New

kubectl-mcp-app - 8 Interactive UI Dashboards

Added a standalone npm package kubectl-mcp-app that provides beautiful, interactive UI dashboards for Kubernetes management using the MCP ext-apps SDK.

Installation:

# Via npm
npm install -g kubectl-mcp-app

# Or via npx (no install)
npx kubectl-mcp-app

Claude Desktop Configuration:

{
  "mcpServers": {
    "kubectl-app": {
      "command": "npx",
      "args": ["kubectl-mcp-app"]
    }
  }
}

8 Interactive UI Tools:

Tool	Description
k8s-pods	Interactive pod viewer with filtering, sorting, status indicators
k8s-logs	Real-time log viewer with syntax highlighting and search
k8s-deploy	Deployment dashboard with rollout status, scaling, rollback
k8s-helm	Helm release manager with upgrade/rollback actions
k8s-cluster	Cluster overview with node health and resource metrics
k8s-cost	Cost analyzer with waste detection and recommendations
k8s-events	Events timeline with type filtering and grouping
k8s-network	Network topology graph showing Services/Pods/Ingress

Features:

• 🎨 Dark/light theme support
• 📊 Real-time data visualization
• 🖱️ Interactive actions (scale, restart, delete)
• 🔗 Seamless integration with kubectl-mcp-server
• 📦 TypeScript + React 19 + Vite
• 🎯 Single-file HTML bundles (~220KB each)
• ✅ 27 tests with full coverage

Full Changelog

v1.21.0...v1.22.0

v1.21.0 - vind, kind & Enhanced Skills

What's New

vCluster (vind) Support - 14 Tools

Manage virtual Kubernetes clusters using vCluster CLI:

• vind_detect_tool, vind_list_clusters_tool, vind_status_tool
• vind_create_cluster_tool, vind_delete_cluster_tool
• vind_pause_tool, vind_resume_tool, vind_connect_tool, vind_disconnect_tool
• vind_upgrade_tool, vind_describe_tool, vind_get_kubeconfig_tool
• vind_logs_tool, vind_platform_start_tool

kind (Kubernetes IN Docker) Support - 32 Tools

Comprehensive local Kubernetes cluster management:

Core Operations (15 tools):

• Cluster lifecycle: create, delete, list, info
• Image management: load, build node images
• Kubeconfig: get, set, export

New in v1.21.0 (17 tools):

• Configuration: kind_config_validate, kind_config_generate, kind_config_show, kind_available_images
• Registry: kind_registry_create, kind_registry_connect, kind_registry_status
• Node Management: kind_node_exec, kind_node_logs, kind_node_inspect, kind_node_restart
• Networking: kind_network_inspect, kind_port_mappings, kind_ingress_setup
• Diagnostics: kind_cluster_status, kind_images_list, kind_provider_info

Enhanced Agent Skills - 26 Skills

All 26 Kubernetes skills updated to follow agenstskills.com specification:

• Enhanced YAML frontmatter (license, author, version, tools, category)
• "When to Apply" activation trigger sections
• Priority rules tables (CRITICAL/HIGH/MEDIUM/LOW)
• Quick reference tables for common operations

New Reference Documents:

• k8s-troubleshoot/references/DECISION-TREE.md, COMMON-ERRORS.md
• k8s-deploy/references/STRATEGIES.md
• k8s-helm/references/CHART-STRUCTURE.md
• k8s-kind/references/CONFIGS.md
• k8s-vind/references/WORKFLOWS.md

New Example Manifests:

• k8s-deploy/examples/canary-rollout.yaml, blue-green.yaml, hpa-deployment.yaml
• k8s-autoscaling/examples/hpa-cpu.yaml, keda-scaledobject.yaml

Tool Counts

• Total Tools: 270+ (was 235+)
• vind tools: 14 (new)
• kind tools: 32 (expanded from 15)
• Agent Skills: 26

Installation

# pip
pip install kubectl-mcp-server==1.21.0

# npm
npx kubectl-mcp-server@1.21.0

# Docker
docker pull rohitghumare64/kubectl-mcp-server:1.21.0

Full Changelog

v1.19.3...v1.21.0

v1.19.3

What's Changed

• Added GitHub Container Registry (ghcr.io) publishing
• Docker images now available on both Docker Hub and GitHub Packages

Docker Images

# Docker Hub
docker pull rohitghumare64/kubectl-mcp-server:1.19.3

# GitHub Container Registry
docker pull ghcr.io/rohitg00/kubectl-mcp-server:1.19.3

Install from GitHub Packages

# npm
npm install @rohitg00/kubectl-mcp-server --registry=https://npm.pkg.github.com

# Python (wheel from release)
pip install https://github.com/rohitg00/kubectl-mcp-server/releases/download/v1.19.3/kubectl_mcp_server-1.19.3-py3-none-any.whl

v1.19.2

What's Changed

• Added GitHub Packages publishing for npm package
• Package now available at @rohitg00/kubectl-mcp-server on GitHub Packages

Install from GitHub Packages

npm install @rohitg00/kubectl-mcp-server --registry=https://npm.pkg.github.com

v1.19.1

What's Changed

• Upload wheel files to GitHub releases for direct pip install
• Updated documentation with GitHub release install instructions

Install from GitHub Release

pip install https://github.com/rohitg00/kubectl-mcp-server/releases/download/v1.19.1/kubectl_mcp_server-1.19.1-py3-none-any.whl

v1.19.0 - Multi-Cluster Support, Kubeconfig Watch & Code Quality

What's New in v1.19.0

New Features

Multi-Cluster Operations (4 new tools)

• multi_cluster_query: Query resources across multiple clusters simultaneously
• multi_cluster_health: Check health status of all clusters at once
• multi_cluster_pod_count: Get pod counts across all clusters with breakdown by phase

Kubeconfig Auto-Detection

• enable_kubeconfig_watching: Auto-detect kubeconfig file changes (useful for cloud CLI credential updates)
• disable_kubeconfig_watching: Disable the watcher
• get_server_config_status: Check current server configuration including watch status

Stateless Mode

• MCP_STATELESS_MODE environment variable: Run without caching API clients (ideal for serverless)
• set_stateless_mode / is_stateless_mode: Toggle stateless operation programmatically

Pod Management

• run_pod: Run container images directly (kubectl run equivalent)

Code Quality Improvements

• Created shared utils.py module, removing ~474 lines of duplicate code
• Removed 31 redundant comments from tool files
• Added input validation for check_interval parameter
• Simplified module docstrings
• All 47 tests passing

Files Changed

• 20 files modified
• New: kubectl_mcp_tool/tools/utils.py

Installation

# PyPI
pip install kubectl-mcp-server==1.19.0

# npm
npm install kubectl-mcp-server@1.19.0

Full Changelog: v1.18.0...v1.19.0

v1.18.0 - Agent Skills Library & Enhanced Provider

What's New

Agent Skills Library (24 Skills)

Added comprehensive Kubernetes skills for AI coding agents following the Agent Skills specification.

Categories:

Category	Skills
Core Resources	k8s-core, k8s-networking, k8s-storage
Workloads	k8s-deploy, k8s-operations, k8s-helm
Observability	k8s-diagnostics, k8s-troubleshoot, k8s-incident
Security	k8s-security, k8s-policy, k8s-certs
GitOps	k8s-gitops, k8s-rollouts
Scaling	k8s-autoscaling, k8s-cost, k8s-backup
Multi-Cluster	k8s-multicluster, k8s-capi, k8s-kubevirt
Networking	k8s-service-mesh, k8s-cilium
Tools	k8s-browser, k8s-cli

Installation:

# Copy all skills to Claude
cp -r kubernetes-skills/claude/* ~/.claude/skills/

# Or convert to other agents with SkillKit
npm install -g skillkit
skillkit translate kubernetes-skills/claude --to cursor --output .cursor/rules/

Enhanced Provider Module

New providers.py module for better multi-cluster management:

• Singleton Pattern: KubernetesProvider.get_instance() for shared provider
• API Client Caching: Cached clients per context for better performance
• Context Validation: UnknownContextError with available contexts
• Environment Variables:
  • MCP_K8S_PROVIDER: kubeconfig, in-cluster, or single
  • MCP_K8S_CONTEXT: Default context for single provider
  • MCP_K8S_QPS: API rate limit (default: 100)
  • MCP_K8S_BURST: API burst limit (default: 200)
  • MCP_K8S_TIMEOUT: Request timeout in seconds (default: 30)

Full Changelog

• 24 new Agent Skills covering all 224+ tools
• New providers.py module with enhanced provider management
• Integrated provider into k8s_config.py with fallback support
• Updated documentation with skills installation guide

Full Changelog: v1.17.0...v1.18.0

v1.17.0 - MCP Server Enterprise Integration

v1.17.0 - MCP Server Enterprise Integration

This release integrates enterprise-readiness modules directly into the MCP server, building on the PRD implementation from previous releases.

MCP Server Integration

HTTP Endpoints Added:

• /stats - Real-time server statistics and tool call metrics
• /metrics - Prometheus-compatible metrics (when available)
• /safety - Safety mode status and configuration

CLI Parameters:

• --config <file> - Load configuration from TOML file
• --read-only - Enable read-only safety mode
• --disable-destructive - Disable destructive operations only

Configuration Features:

• TOML configuration file support
• SIGHUP signal handler for runtime config reload
• CLI flags take precedence over config file settings
• Custom prompts loading from config

Safety Modes

• Normal: All operations allowed
• Read-Only: Only read operations (get, list, describe)
• Disable-Destructive: Block destructive operations (delete, drain, cordon)

Observability

• StatsCollector for tool call metrics
• Prometheus metrics integration (optional)
• OpenTelemetry tracing support (optional)

Example Usage

# Start with config file
kubectl-mcp-server serve --config /etc/kubectl-mcp/config.toml

# Start in read-only mode
kubectl-mcp-server serve --read-only

# Disable destructive operations only
kubectl-mcp-server serve --disable-destructive

Installation

• pip: pip install kubectl-mcp-server==1.17.0
• npm: npx kubectl-mcp-server@1.17.0
• Docker: docker pull rohitghumare64/kubectl-mcp-server:1.17.0

Full Changelog

v1.16.0...v1.17.0