cloudthinker-ai Kubecost Management Skill
Manage and monitor Kubernetes cost allocation and optimization with Kubecost.
MANDATORY: Discovery-First Pattern
Always check Kubecost availability and cluster coverage before querying costs.
Phase 1: Discovery
#!/bin/bash
kubecost_api() {
local endpoint="$1"
curl -s "http://${KUBECOST_HOST:-localhost:9090}/model/${endpoint}"
}
echo "=== Kubecost Status ==="
kubecost_api "status" | jq '{
version: .version,
clusterCount: .clusterCount,
dataStatus: .dataStatus
}' 2>/dev/null || echo "Checking Kubecost pod..."
kubectl get pods -A -l app=cost-analyzer 2>/dev/null | head -5
echo ""
echo "=== Monitored Clusters ==="
kubecost_api "clusterInfo" | jq -r '.[] | "\(.id)\t\(.name)\t\(.provider)"' | column -t
echo ""
echo "=== Namespace Cost Summary (last 24h) ==="
kubecost_api "allocation?window=1d&aggregate=namespace" | jq -r '
.data[0] | to_entries[] |
"\(.key)\t$\(.value.totalCost | . * 100 | round / 100)\tCPU: $\(.value.cpuCost | . * 100 | round / 100)\tRAM: $\(.value.ramCost | . * 100 | round / 100)"
' | sort -t'$' -k2 -rn | column -t | head -20
Core Helper Functions
#!/bin/bash
kubecost_api() {
local endpoint="$1"
curl -s "http://${KUBECOST_HOST:-localhost:9090}/model/${endpoint}"
}
kubecost_allocation() {
local window="${1:-1d}"
local aggregate="${2:-namespace}"
kubecost_api "allocation?window=${window}&aggregate=${aggregate}"
}
kubecost_savings() {
kubecost_api "savings"
}
Output Rules
- TOKEN EFFICIENCY: Target ≤50 lines per output
- Use jq filtering on API responses to extract key cost fields
- Round cost values to 2 decimal places for readability
Common Operations
Namespace Cost Breakdown
#!/bin/bash
WINDOW="${1:-7d}"
echo "=== Namespace Costs (window: $WINDOW) ==="
kubecost_api "allocation?window=${WINDOW}&aggregate=namespace" | jq -r '
.data[0] | to_entries[] |
"\(.key)\t$\(.value.totalCost | . * 100 | round / 100)\tCPU: \(.value.cpuCoreRequestAverage | . * 100 | round / 100) cores\tRAM: \(.value.ramByteRequestAverage / 1073741824 | . * 100 | round / 100) GiB"
' | sort -t'$' -k2 -rn | column -t | head -20
echo ""
echo "=== Total Cluster Cost ==="
kubecost_api "allocation?window=${WINDOW}&aggregate=cluster" | jq -r '
.data[0] | to_entries[] |
"\(.key)\tTotal: $\(.value.totalCost | . * 100 | round / 100)"
' | column -t
Workload Efficiency Scoring
#!/bin/bash
NAMESPACE="${1:-}"
WINDOW="${2:-2d}"
FILTER=""
[ -n "$NAMESPACE" ] && FILTER="&filterNamespaces=${NAMESPACE}"
echo "=== Workload Efficiency (window: $WINDOW) ==="
kubecost_api "allocation?window=${WINDOW}&aggregate=controller${FILTER}" | jq -r '
.data[0] | to_entries[] |
select(.value.totalCost > 0.01) |
"\(.key)\tCost: $\(.value.totalCost | . * 100 | round / 100)\tCPU Eff: \((.value.cpuEfficiency // 0) * 100 | round)%\tRAM Eff: \((.value.ramEfficiency // 0) * 100 | round)%\tTotal Eff: \((.value.totalEfficiency // 0) * 100 | round)%"
' | sort -t'$' -k2 -rn | column -t | head -20
Savings Recommendations
#!/bin/bash
echo "=== Right-Sizing Recommendations ==="
kubecost_api "savings/requestSizing?window=48h&targetCPUUtilization=0.65&targetRAMUtilization=0.70" | jq -r '
.[] | select(.annualSavings > 10) |
"\(.namespace)/\(.controllerName)\tSavings: $\(.annualSavings | round)/yr\tCPU: \(.currentCPURequest)→\(.recommendedCPURequest)\tRAM: \(.currentRAMRequest)→\(.recommendedRAMRequest)"
' | sort -t'$' -k2 -rn | head -15
echo ""
echo "=== Abandoned Workloads ==="
kubecost_api "savings/abandonedWorkloads?window=7d" | jq -r '
.[] | "\(.namespace)/\(.name)\tLast Active: \(.lastSeen[0:16])\tCost: $\(.monthlyCost | . * 100 | round / 100)/mo"
' | head -10
Cost Trend Analysis
#!/bin/bash
echo "=== Daily Cost Trend (last 7 days) ==="
kubecost_api "allocation?window=7d&aggregate=cluster&step=1d" | jq -r '
.data[] | to_entries[] |
"\(.value.start[0:10])\t$\(.value.totalCost | . * 100 | round / 100)"
' | column -t
echo ""
echo "=== Cost by Label (team) ==="
kubecost_api "allocation?window=7d&aggregate=label:team" | jq -r '
.data[0] | to_entries[] |
select(.key != "__unallocated__") |
"\(.key)\t$\(.value.totalCost | . * 100 | round / 100)"
' | sort -t'$' -k2 -rn | column -t | head -15
Asset Cost Analysis
#!/bin/bash
echo "=== Asset Costs by Type (last 7d) ==="
kubecost_api "assets?window=7d&aggregate=type" | jq -r '
.data[0] | to_entries[] |
"\(.key)\t$\(.value.totalCost | . * 100 | round / 100)"
' | sort -t'$' -k2 -rn | column -t
echo ""
echo "=== Node Costs ==="
kubecost_api "assets?window=7d&aggregate=node" | jq -r '
.data[0] | to_entries[] |
"\(.key)\t$\(.value.totalCost | . * 100 | round / 100)\tType: \(.value.properties.instanceType // "unknown")"
' | sort -t'$' -k2 -rn | column -t | head -15
Safety Rules
- Read-only by default: Kubecost is primarily observational -- no destructive operations
- Window validation: Ensure window parameter matches data retention period
- Cost accuracy: Kubecost estimates may differ from cloud billing -- use for relative comparison
- Efficiency thresholds: Do not auto-resize based solely on efficiency scores -- validate with team
Output Format
Present results as a structured report:
Managing Kubecost Report
════════════════════════
Resources discovered: [count]
Resource Status Key Metric Issues
──────────────────────────────────────────────
[name] [ok/warn] [value] [findings]
Summary: [total] resources | [ok] healthy | [warn] warnings | [crit] critical
Action Items: [list of prioritized findings]
Target ≤50 lines of output. Use tables for multi-resource comparisons.
Anti-Hallucination Rules
- NEVER assume resource names — always discover via CLI/API in Phase 1 before referencing in Phase 2.
- NEVER fabricate metric names or dimensions — verify against the service documentation or
--helpoutput. - NEVER mix CLI commands between service versions — confirm which version/API you are targeting.
- ALWAYS use the discovery → verify → analyze chain — every resource referenced must have been discovered first.
- ALWAYS handle empty results gracefully — an empty response is valid data, not an error to retry.
Counter-Rationalizations
| Shortcut | Counter | Why |
|---|---|---|
| "I'll skip discovery and check known resources" | Always run Phase 1 discovery first | Resource names change, new resources appear — assumed names cause errors |
| "The user only asked for a quick check" | Follow the full discovery → analysis flow | Quick checks miss critical issues; structured analysis catches silent failures |
| "Default configuration is probably fine" | Audit configuration explicitly | Defaults often leave logging, security, and optimization features disabled |
| "Metrics aren't needed for this" | Always check relevant metrics when available | API/CLI responses show current state; metrics reveal trends and intermittent issues |
| "I don't have access to that" | Try the command and report the actual error | Assumed permission failures prevent useful investigation; actual errors are informative |
Common Pitfalls
- Data lag: Kubecost needs 24-48h of data before efficiency scores stabilize
- Shared costs: Cluster overhead (system pods, control plane) may not be allocated to namespaces
- Idle costs: Unallocated resources appear as "idle" -- this is not waste, it is capacity headroom
- Network costs: Network egress costs are often estimated, not precise
- Multi-cluster: Ensure the correct cluster context when querying multi-cluster Kubecost
- Prometheus dependency: Kubecost relies on Prometheus/Thanos -- if metrics are missing, costs will be incomplete