k13d Feature Gap Analysis¶

Overview¶

This document compares k13d with three major Kubernetes dashboard/AI projects to identify missing features.

Analyzed Projects: - kubernetes-dashboard - Official Kubernetes Dashboard - headlamp - Cloud-native Kubernetes UI with plugin system - kubectl-ai - AI-powered kubectl assistant

Last Updated: 2026-01-22

Current k13d Features (Baseline)¶

CLI/TUI Features¶

• Resource listing: Pods, Deployments, Services, Nodes, Namespaces, Events, ConfigMaps, Secrets, DaemonSets, StatefulSets, Jobs, CronJobs, Ingresses
• Namespace switching
• Context switching
• Real-time table filtering with highlight
• Command autocomplete with hints
• AI chat integration (streaming)
• YAML view for resources
• Pod logs viewer
• Delete with confirmation
• Port forwarding
• Shell exec (spawns external terminal)
• Health check command
• i18n support (en, ko, ja, zh)

Web UI Features (index.html)¶

• Login page with username/password
• Dark theme (Tokyo Night style)
• Sidebar navigation with resource groups
• Resource table with status colors
• Namespace selector dropdown
• AI chat panel with streaming responses
• Resizable panels (drag handle)
• Settings modal (language, LLM config)
• Audit log viewer
• Reports viewer
• Refresh button per resource
• User badge and logout
• Tab system (Table/YAML/Logs)
• Message history in AI panel
• Tool execution visualization with Show More button

Web Server/API Features¶

• REST API for resources (pods, deployments, services, namespaces, nodes, events)
• AI chat endpoint (SSE streaming with agentic mode)
• JWT authentication with session management
• LDAP authentication support
• Audit logging (SQLite)
• Report generation API
• Settings API (GET/PUT)
• LLM settings API
• Health check endpoint
• CORS middleware
• Static file serving (embedded)
• Tool approval flow for write/dangerous commands
• Command classification (read-only, write, dangerous)

Backend Features¶

• Multi-provider LLM support (OpenAI-compatible, Ollama, Anthropic)
• Kubernetes client with metrics API
• Dynamic client for CRDs
• Context switching support
• MCP (Model Context Protocol) integration
• Tool registry with MCP tool support

Recently Implemented Features (2026-01)¶

AI Safety & Validation ✅¶

Status: Implemented - [x] Command classification (read-only, write, dangerous) - [x] Tool approval flow for destructive operations - [x] Auto-approve read-only commands - [x] User confirmation required for write/dangerous commands - [x] Timeout handling for approval requests

In-Browser Terminal ✅¶

Status: Implemented - [x] WebSocket-based terminal (pkg/web/terminal.go) - [x] Bidirectional stdin/stdout streaming - [x] Terminal resize support - [x] Container selection support

Deployment Operations ✅¶

Status: Implemented (pkg/web/operations.go) - [x] Scale deployments - [x] Rollout restart - [x] Pause/resume deployment - [x] Rollback to previous revision - [x] View revision history

StatefulSet/DaemonSet Operations ✅¶

Status: Implemented - [x] Scale StatefulSets - [x] Restart StatefulSets - [x] Restart DaemonSets

CronJob Operations ✅¶

Status: Implemented - [x] Trigger CronJob manually (create Job) - [x] Suspend/resume CronJob

Node Operations ✅¶

Status: Implemented - [x] Node cordon/uncordon - [x] Node drain with eviction - [x] List pods on node

Helm Integration ✅¶

Status: Implemented (pkg/helm/client.go) - [x] List Helm releases - [x] Get release details - [x] View release history - [x] Get release values - [x] Get release manifest - [x] Install releases - [x] Upgrade releases - [x] Uninstall releases - [x] Rollback releases - [x] Repository management (add, remove, update) - [x] Search charts

MCP Support ✅¶

Status: Implemented - [x] MCP client mode (use external tools) - [x] Tool registration system - [x] MCP server connection management - [x] Dynamic tool discovery

Priority 1: Critical Missing Features¶

1.1 Multi-Provider LLM Support (from kubectl-ai)¶

Status: ✅ Implemented (2026-01-22) - [x] OpenAI-compatible provider (with tool calling) - [x] Ollama (local) provider - [x] Model profile management - [x] Google Gemini provider - [x] AWS Bedrock provider (Claude) - with tool calling - [x] Azure OpenAI provider - with tool calling - [ ] Llama.cpp provider (low priority) - [ ] Grok provider (low priority)

1.2 Plugin System (from headlamp)¶

Status: Not Implemented Required: - [ ] Plugin registry system - [ ] Dynamic plugin loading from URL - [ ] Plugin configuration management - [ ] Extension points: - Custom sidebar items - Custom routes - Custom resource detail views - Custom table columns - Custom themes

Files to create:

pkg/plugins/
├── registry.go      # Plugin registry
├── loader.go        # Dynamic plugin loading
├── config.go        # Plugin configuration
└── types.go         # Plugin interfaces

Priority 2: Important Missing Features¶

2.1 Session Persistence (from kubectl-ai)¶

Status: ✅ Implemented (2026-01-22) Location: pkg/ai/session/session.go - [x] Session storage interface (Store) - [x] Filesystem backend (JSON files in XDG data directory) - [x] Session metadata (ID, timestamps, model, provider) - [x] Resume/load previous sessions (Get, GetContextMessages) - [x] Clear conversation command (Clear, Delete) - [x] List sessions with pagination (List, GetRecentSessions) - [x] Export/Import sessions - [x] Tool call recording in messages

2.2 Resource Graph Visualization (from headlamp)¶

Status: Not Implemented Required: - [ ] Pod-to-Service-to-Deployment graph - [ ] Resource relationship mapping - [ ] Interactive graph navigation - [ ] Namespace grouping

Implementation: Use xyflow/react or similar for web UI

2.3 Metrics Visualization (from kubernetes-dashboard)¶

Status: API Available, UI Not Implemented Required: - [ ] CPU/Memory sparklines in table - [ ] Resource usage charts - [ ] Historical metrics storage - [ ] Metrics aggregation

Priority 3: Enhancement Features¶

3.1 OIDC/OAuth2 Authentication (from headlamp)¶

Status: Not Implemented Required: - [ ] OIDC provider integration - [ ] PKCE flow support - [ ] Token refresh automation - [ ] JMESPath claim extraction - [ ] Multiple auth method support

3.2 Multi-Cluster Support (from headlamp)¶

Status: Not Implemented Required: - [ ] Cluster switcher UI - [ ] Per-cluster authentication - [ ] Kubeconfig file watcher - [ ] Dynamic cluster addition

3.3 Advanced Search (from headlamp)¶

Status: Not Implemented Required: - [ ] Global search across clusters - [ ] Label selector filtering - [ ] Field selector filtering - [ ] Recent searches tracking - [ ] Full-text search with fuse.js

3.4 Custom Resource Definition Support (from kubernetes-dashboard, headlamp)¶

Status: Partial (API exists) Required: - [ ] List all CRDs in UI - [ ] Browse CRD instances - [ ] Create/edit custom resources - [ ] CRD schema validation - [ ] OpenAPI documentation display

3.5 Retry Logic with Backoff (from kubectl-ai)¶

Status: ✅ Implemented Location: pkg/ai/providers/factory.go - [x] Exponential backoff - [x] Jitter support - [x] Retryable error detection (429, 5xx, timeout, connection errors) - [x] Max attempts configuration - [x] Configurable via RetryConfig

3.6 YAML Editor with Validation (from headlamp)¶

Status: Not Implemented Required: - [ ] Monaco editor integration - [ ] Syntax highlighting - [ ] Schema validation - [ ] Auto-completion - [ ] Inline documentation

Priority 4: UI/UX Improvements¶

4.1 From kubernetes-dashboard¶

• Deployment creation wizard (form-based)
• Image reference validation
• Protocol validation for services
• Log download as file
• Multi-container log selection
• Sparkline metrics in list views
• CSRF protection for API

4.2 From headlamp¶

• Dark/Light theme toggle
• Custom theme support
• Responsive mobile layout
• Breadcrumb navigation
• Action confirmation dialogs
• Empty state handling
• Loading animations
• Error boundary UI
• Internationalization (12 languages)

4.3 From kubectl-ai¶

• Streaming log UI for AI responses
• Meta-commands (clear, model, tools, sessions)
• Tool call visualization
• Audit/journal logging with structured output

API Feature Gaps (Backend)¶

From kubernetes-dashboard API¶

From headlamp API¶

Summary¶

Implemented Features: ~30 major features Remaining Features: ~15 major features

Implementation Progress¶

Completed (Phase 1-2)¶

1. ✅ AI safety validation with tool approval flow
2. ✅ In-browser terminal (WebSocket)
3. ✅ Deployment operations (scale, restart, pause, resume, rollback, history)
4. ✅ StatefulSet/DaemonSet operations
5. ✅ CronJob operations (trigger, suspend)
6. ✅ Node operations (cordon, drain, pods)
7. ✅ Helm integration (releases, repos, search)
8. ✅ MCP support
9. ✅ Tool execution visualization

Recently Completed (2026-01-22)¶

• Google Gemini provider
• AWS Bedrock provider (Claude) with tool calling
• Azure OpenAI provider with tool calling
• Session persistence for AI conversations
• Retry logic with exponential backoff

Upcoming (Phase 3-5)¶

1. Metrics visualization (sparklines, charts)
2. Resource graph visualization
3. Plugin system
4. Multi-cluster support
5. OIDC authentication
6. Advanced search
7. YAML editor with Monaco
8. Web UI integration for session management