Kubernetes Cost Prediction Action

Predict the cost of Kubernetes manifests (specs) in CI! Make cost decisions before merging changes.

This is a GitHub Action, powered by Kubecost, to make cost predictions for K8s workloads before they are applied to your cluster. It does not require you to have Kubecost installed, but will produce highly-accurate, environment-specific predictions if you do.

In action:

---

Actions in this Repository

Action	Description
kubecost/cost-prediction-action	Base action — predicts cost of K8s manifests and outputs a formatted table
./action-enhanced	Enhanced action — adds delta/diff mode, cost threshold enforcement, and budget checks on top of the base action

---

Base Action Usage

Add this Action as a step in one of your Actions workflows and point it at a single YAML file or a directory containing at least one YAML file. Non-YAML files will be ignored. The YAML files will be interpreted as Kubernetes manifests and a cost prediction will be run on supported Kubernetes object types.

If you aren't familiar with GitHub Actions, check out GitHub's quickstart documentation.

Simple

- name: Run prediction
  id: prediction
  uses: kubecost/cost-prediction-action@v0.1.1
  with:
    path: ./repo

- name: Find existing PR comment
  uses: kubecost/github-actions/find-comment@main
  id: find-comment
  with:
    token: ${{ secrets.GITHUB_TOKEN }}
    issue-number: ${{ github.event.pull_request.number }}
    comment-author: 'github-actions[bot]'
    body-includes: '<!-- kubecost-prediction-results -->'
- name: Create or update PR comment with prediction results
  uses: kubecost/github-actions/create-or-update-comment@main
  with:
    token: ${{ secrets.GITHUB_TOKEN }}
    issue-number: ${{ github.event.pull_request.number }}
    comment-id: ${{ steps.find-comment.outputs.comment-id }}
    edit-mode: replace
    body: |
      <!-- kubecost-prediction-results -->

      ## Kubecost's total cost prediction for K8s YAML Manifests in this PR

      \```
      ${{ steps.prediction.outputs.PREDICTION_TABLE }}
      \```

Advanced (full workflow)

name: Predict K8s spec cost
on: [pull_request]

jobs:
  predict-cost:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          path: ./repo

      # Uncomment for GKE / port-forwarded Kubecost access:
      # - name: Forward the kubecost service
      #   run: |
      #     kubectl port-forward --namespace kubecost service/kubecost-cost-analyzer 9090 &
      #     sleep 5

      # For Helm-based repos, template first:
      # - name: Helm template
      #   run: helm template RELEASENAME ./repo >> ./templated.yaml

      - name: Run prediction
        id: prediction
        uses: kubecost/cost-prediction-action@v0.1.1
        with:
          log_level: "info"
          path: ./repo
          # kubecost_api_path: "http://localhost:9090/model"

      - name: Find existing PR comment
        uses: kubecost/github-actions/find-comment@main
        id: find-comment
        with:
          token: ${{ secrets.GITHUB_TOKEN }}
          issue-number: ${{ github.event.pull_request.number }}
          comment-author: 'github-actions[bot]'
          body-includes: '<!-- kubecost-prediction-results -->'
      - name: Create or update PR comment with prediction results
        uses: kubecost/github-actions/create-or-update-comment@main
        with:
          token: ${{ secrets.GITHUB_TOKEN }}
          issue-number: ${{ github.event.pull_request.number }}
          comment-id: ${{ steps.find-comment.outputs.comment-id }}
          edit-mode: replace
          body: |
            <!-- kubecost-prediction-results -->

            ## Kubecost's total cost prediction for K8s YAML Manifests in this PR

            \```
            ${{ steps.prediction.outputs.PREDICTION_TABLE }}
            \```

Base Action Inputs/Outputs

Inputs

Name	Description	Required	Default
path	Path to a file or directory of K8s YAML manifests	Yes
kubecost_api_path	URL of your Kubecost API (e.g. https://kubecost.example.com:9090/model). If omitted, default pricing is used.	No
log_level	Log verbosity: debug, info, warn, error	No	info

Outputs

Name	Description
PREDICTION_TABLE	ASCII-formatted cost prediction table

---

Enhanced Action Usage

The enhanced action (./action-enhanced) wraps the base action and adds:

1. Delta/Diff Mode — compares cost between your PR branch and a base branch
2. Cost Threshold Enforcement — fails the workflow if cost increases exceed configured limits
3. Budget Checks — validates predicted cost against a Kubecost budget

Example: Full Governance Workflow

name: Cost Governance
on: [pull_request]

jobs:
  cost-check:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
        with:
          fetch-depth: 0  # Required for delta mode

      - name: Check cost impact
        id: cost
        uses: kubecost/cost-prediction-action/action-enhanced@main
        with:
          path: ./k8s
          kubecost_api_path: ${{ secrets.KUBECOST_API_PATH }}
          enable_delta_mode: "true"
          base_ref: "main"
          max_cost_increase: "100.00"
          max_cost_percentage: "15"
          fail_on_threshold: "true"
          budget_check_enabled: "true"
          budget_id: "team-alpha-budget"
          fail_on_budget_exceeded: "true"

      - name: Comment on PR
        uses: actions/github-script@v7
        with:
          script: |
            github.rest.issues.createComment({
              issue_number: context.issue.number,
              owner: context.repo.owner,
              repo: context.repo.repo,
              body: `## 💰 Cost Impact Analysis
              | | |
              |---|---|
              | **Current Cost** | \`$${{ steps.cost.outputs.TOTAL_MONTHLY_COST }}/mo\` |
              | **Base Cost** | \`$${{ steps.cost.outputs.BASE_COST }}/mo\` |
              | **Change** | \`${{ steps.cost.outputs.COST_CHANGE }}\` (\`${{ steps.cost.outputs.COST_CHANGE_PERCENTAGE }}%\`) |
              | **Budget Status** | \`${{ steps.cost.outputs.BUDGET_STATUS }}\` |
              | **Budget Remaining** | \`$${{ steps.cost.outputs.BUDGET_REMAINING }}\` |`
            });

Enhanced Action Inputs

Base Inputs (inherited)

Input	Description	Default
path	Path to workload file or directory	(required)
kubecost_api_path	URL of Kubecost API
log_level	Log level	info

Delta Mode

Input	Description	Default
enable_delta_mode	Compare costs between current and base branch	false
base_ref	Base branch to compare against	main

Cost Thresholds

Input	Description	Default
max_cost_increase	Max allowed cost increase in dollars	(none)
max_cost_percentage	Max allowed cost increase percentage	(none)
fail_on_threshold	Fail workflow when threshold exceeded	true

Budget Checks

Input	Description	Default
budget_check_enabled	Enable budget validation	false
budget_id	Kubecost budget ID to check against	(none)
fail_on_budget_exceeded	Fail workflow when budget would be exceeded	true

Enhanced Action Outputs

Output	Description
PREDICTION_TABLE	ASCII cost prediction table
PREDICTION_JSON	JSON-formatted prediction data
TOTAL_MONTHLY_COST	Total monthly cost
BASE_COST	Base branch cost (delta mode only)
COST_CHANGE	Absolute cost difference (delta mode only)
COST_CHANGE_PERCENTAGE	Percentage cost change (delta mode only)
THRESHOLD_EXCEEDED	Whether thresholds were exceeded (true/false)
BUDGET_STATUS	WITHIN_BUDGET, EXCEEDS_BUDGET, or NOT_CHECKED
BUDGET_TOTAL	Budget spend limit in dollars
BUDGET_CURRENT_SPEND	Current spend against the budget
BUDGET_REMAINING	Remaining budget before this deployment
BUDGET_UTILIZATION_PERCENTAGE	Current utilization % before this deployment

---

Limitations

• Only .yml/.yaml manifest files are supported. For Helm, run helm template first.
• A limited set of Kubernetes object types are supported. More are planned.
• Manifests without container resource requests will not produce predictions.
• Delta mode requires fetch-depth: 0 in the checkout step so that the base branch is accessible.

---

Development

Source code for the prediction container is in the upstream Kubecost repo. Kubecost engineers: see cmd/costpredictionaction in KCM for development, testing, and releasing details.