FishKube Manager

FishKube Manager is a modern, self-hosted Kubernetes management platform with multi-cluster support, AI-powered operations, fine-grained RBAC, and a mobile-friendly UI.

FishKube Manager is a modern SAAS Kubernetes management dashboard, featuring frontend and a Helm chart for Kubernetes deployment.

It focuses on operational visibility, safe resource management, and optional AI assisted cluster interactions.

Built for real-world DevOps — triage incidents from your phone, manage clusters from a single pane of glass, and let AI handle the repetitive queries.

Features

• Cluster or namespace overview dashboard
• Resource browsing for core Kubernetes objects
• Modern logs viewer
• YAML viewer and editor with RBAC
• Local and Google OAuth authentication
• Optional AI assistant for cluster operations
• Optional external monitoring links (Grafana)

Screenshots

• Dashboard overview
• Pods table and resource usage
• Logs viewer
• YAML editor
• Chat AI assistant

Full Feature List

• Multi-cluster management — local, kubeconfig, and agent-based cluster connections in one UI
• Cluster overview dashboard — nodes, pods, namespaces, CPU/memory usage at a glance
• Resource browser — Pods, Deployments, DaemonSets, StatefulSets, ReplicaSets, Jobs, CronJobs, Services, Ingresses, ConfigMaps, Secrets, PVCs, PVs, Nodes, Events
• Logs viewer — real-time pod logs with ANSI color support, filtering, and wrap toggle
• YAML viewer & editor — syntax-highlighted YAML with inline editing and server-side apply
• Scale operations — scale deployments up/down from the UI or via natural language in Chat AI
• AI Chat assistant — ask questions in plain English, get answers from live cluster data
  • Multi-provider: OpenAI, Google Gemini, AWS Bedrock (Claude)
  • Cross-cluster queries across all connected clusters
  • Session memory and suggested follow-up questions
  • Safe by design — write operations require explicit confirmation
• Authentication — local user database or Google SSO (or both)
• Permission groups — Administrators, DevOps, Team Leads, RnD, Read Only (+ custom groups)
• Per-cluster permissions — assign different permission groups per user per cluster
• Default cluster access — set a baseline permission group per cluster for all users
• Custom cluster colors — visual cluster identification with color-coded selectors
• Jira integration — ticket-gated tier changes with API verification, AI content validation, and auto-commenting
• Slack notifications — real-time alerts for tier changes and scale operations
• Prometheus integration — 1h average CPU/memory usage per pod
• External monitoring deep-links — configurable links to Grafana, Datadog, or any monitoring tool
• Mobile-responsive UI — fully functional on phones and tablets
• Dark/light theme — toggle from the header

Screenshots

Architecture

A single Helm chart deploys both the manager (full UI) and the agent (lightweight proxy for remote clusters):

Mode	What it deploys	Use case
manager	Full management UI with auth, AI, SQLite	Central control plane
agent	Lightweight Kubernetes API proxy	Remote clusters behind firewalls

Connection types:

Type	How it works	Best for
Local	In-cluster ServiceAccount	The cluster where the manager runs
Kubeconfig	Stored kubeconfig YAML	Remote clusters with direct API access
Agent	HTTP proxy deployed on remote cluster	Clusters behind firewalls or different VPCs

Quick Start

# Create namespace and admin secret
kubectl create namespace fishkube-manager
kubectl create secret generic fishkube-admin \
  --namespace fishkube-manager \
  --from-literal=username=admin \
  --from-literal=password='your-secure-password'

# Deploy the manager
helm upgrade --install fishkube-manager ./helm \
  -n fishkube-manager -f helm/values.yaml

# Access the UI
kubectl port-forward -n fishkube-manager svc/fishkube-manager 8080:80
open http://localhost:8080

Connect a Remote Cluster (Agent)

# On the remote cluster
helm upgrade --install fishkube-agent ./helm \
  -f helm/values-agent.yaml \
  -n fishkube-manager --create-namespace \
  --set agent.token='shared-secret-token'

Then register the agent in Admin > Clusters with its URL and token.

Configuration

Environment Variables

Variable	Default	Description
FKM_SECRET	—	Flask session secret key
FKM_DB	/data/users.db	SQLite database path
FKM_BOOTSTRAP_USER	admin	Initial admin username
FKM_BOOTSTRAP_PASS	change_me	Initial admin password
FKM_AI_PROVIDER	openai	AI provider: openai, gemini, or bedrock_anthropic
FKM_SCALE_WAIT	45	Max wait seconds for scale operations
FKM_CHAT_BUDGET	120	Max request budget seconds for AI Chat
FKM_AGENT_TIMEOUT	180	HTTP timeout for agent proxy requests
FKM_PROM_URL	—	Prometheus URL for CPU/memory metrics
FKM_MONITORING_BASE_URL	—	Monitoring tool base URL for deep-links
FKM_MONITORING_QUERY_TEMPLATE	—	URL template with <namespace> placeholder
FKM_EXTERNAL_MONITORING_URL_BTN	false	Show monitoring deep-link button
FKM_AUTH_PROVIDER	local	Auth provider: local or google

AI Providers

OpenAI: OPENAI_API_KEY, OPENAI_MODEL, OPENAI_TEMPERATURE

Google Gemini: GEMINI_API_KEY, GEMINI_MODEL

AWS Bedrock: AWS_REGION, BEDROCK_MODEL_ID (requires bedrock:InvokeModel IAM permission)

Integrations

Key	Description
integrations.jira.baseUrl	Jira instance URL
integrations.jira.userEmail	Jira API user email
integrations.jira.apiToken	Jira API token
integrations.slack.botToken	Slack Bot OAuth token
integrations.slack.channelId	Slack channel ID for notifications

Leave empty to disable. The system works fully without Jira/Slack.

Helm Values

See helm/values.yaml for all defaults. Key sections:

• mode — manager or agent
• authProvider — local vs Google SSO
• persistence — PVC size, mount path, storage class
• ai — provider selection and credentials
• integrations — Jira and Slack configuration
• env — operational environment variables

Security Notes

• Do not commit API keys or secrets to git
• Use Kubernetes Secrets for sensitive values
• Always set a strong agent.token for agent connections
• Set FKM_SECRET to a strong random value in production

Permission Groups

Group	Description
Administrators	Full platform control including user management
DevOps	All resource operations except user management
Team Leads	View all, scale workloads, view logs
RnD	View all, view logs, delete pods
Read Only	View all resources and logs, no mutations

Custom groups can be created with fine-grained permission control. Per-cluster access allows different groups per user per cluster. Default cluster access groups provide baseline permissions for all users.

Run Locally

python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt

export FKM_DB="$(pwd)/data/users.db"
export FKM_SECRET="change-me"
export FKM_BOOTSTRAP_USER="admin"
export FKM_BOOTSTRAP_PASS="change_me"

python -c "from app.app import app; app.run(host='0.0.0.0', port=8080, debug=True)"

License

Apache License 2.0