Documentation
Get started in 60 seconds
One Helm command. Full AI-powered incident intelligence on your cluster.
01 Install with Helm
Prerequisites: Kubernetes 1.25+ · Helm 3.10+ · kubectl
terminal — helm install
# Add the Tagent Helm repository
helm repo add tagent https://tagent-dev.github.io/Tagent
helm repo update
# Install Tagent (accessible on port 7777)
helm install tagent tagent/tagent --namespace tagent --create-namespace
# Verify all pods are running
kubectl get pods -n tagent
# Get access URL
kubectl get svc tagent-web -n tagent
→ Open http://<NodeIP>:7777 in your browser
02 Cloud Access (LoadBalancer)
For AWS/GCP/Azure — always accessible, no port-forward needed.
terminal — cloud deploy
helm install tagent tagent/tagent --namespace tagent --create-namespace \
--set web.service.type=LoadBalancer
# Wait 1-2 minutes for cloud provider to assign IP
kubectl get svc tagent-web -n tagent -w
→ Open the EXTERNAL-IP shown — dashboard is always accessible
03 Install the CLI
terminal — mac/linux
curl -sSL https://raw.githubusercontent.com/Tagent-dev/Tagent/main/scripts/install.sh | sh
# Set your API URL
export TAGENT_API_URL=http://<your-cluster-ip>:8080
powershell — windows
git clone https://github.com/Tagent-dev/Tagent.git
cd Tagent\cli
go build -o tagent.exe .
04 CLI Commands
terminal — tagent cli
# Cluster overview
tagent status
# List all active incidents
tagent incidents
# List all pods with status
tagent pods
# List all nodes with health
tagent nodes
# Ask AI about your cluster (powered by local Ollama)
tagent chat "how many pods are running?"
tagent chat "why is checkout-api crashing?"
tagent chat "what changed before the incident?"
# Analyze a specific incident
tagent analyze INC-0001
# Execute remediation
tagent remediate restart-pod production/pod-name
tagent remediate restart-pod production/pod-name --dry-run
tagent remediate scale-deployment production/checkout-api
# Night Guardian status (autonomous overnight fixes)
tagent guardian
# Health check
tagent health
# Version
tagent version
05 What Gets Deployed
Web DashboardReal-time cluster monitoring UI (port 7777)
API GatewayRoutes all API requests internally
DiscoveryScans Kubernetes resources every 15s
MonitoringDetects CrashLoopBackOff, OOMKilled, Node NotReady
AI EngineLocal LLM chat — Ollama with llama3.1:8b
RemediationAuto-fix engine with Night Guardian mode
NotificationSlack + email alerts with escalation chain
OllamaLocal AI model runtime (no cloud APIs)
PostgreSQLStores incidents, reports, audit logs
06 Configuration
terminal — full config
helm install tagent tagent/tagent --namespace tagent --create-namespace \
--set web.service.type=LoadBalancer \
--set nightGuardian.enabled=true \
--set remediation.mode=auto \
--set secrets.slackWebhookUrl="https://hooks.slack.com/..." \
--set secrets.smtpHost="smtp.gmail.com" \
--set secrets.smtpUser="you@gmail.com" \
--set secrets.smtpPassword="app-password" \
--set notification.smtpTo="oncall@company.com"
07 Upgrade & Uninstall
terminal — lifecycle
# Upgrade to latest
helm repo update
helm upgrade tagent tagent/tagent -n tagent
# Uninstall completely
helm uninstall tagent -n tagent
kubectl delete namespace tagent
08 Key Features
→Local AI Only — Runs entirely on your hardware. No data leaves your cluster.
→Auto-Detection — Monitors pods, nodes, deployments in real-time.
→Night Guardian — Autonomous overnight remediation with configurable confidence.
→Escalation Chain — Slack → Email → Phone call → Auto-fix.
→Video Briefing — Morning AI briefing explaining what happened overnight.
→Incident Reports — Auto-generated postmortems.
→Knowledge Base — Learns from past incidents.
→CLI — Full terminal access to all features.