Developer Tools
Kubernetes YAML Validator & Linter
Validate Kubernetes manifests in your browser — YAML syntax, required fields, current apiVersions, selector/label mismatches, image tags, resource limits, probes and security settings.
How to use
- 1Paste one or more manifests (separated by ---).
- 2Fix the errors, and review the best-practice warnings.
How it's calculated
Checks YAML syntax, apiVersion/kind/metadata.name, RFC 1123 names, deprecated apiVersions, Deployment selector ↔ template label match, container names/images/ports, resources, probes (strict mode), privileged containers, Service ports/selectors and Secret base64.
Frequently asked questions
Is this the same as kubectl --dry-run?
No — it checks common mistakes offline without a cluster. kubectl --dry-run=server also validates against your cluster's exact API schema and admission rules.
Why warn about :latest images?
Untagged or latest images make deployments non-reproducible and can pull unexpected versions on restart.
Are my manifests uploaded?
No — validation happens in your browser, so secrets in your YAML stay local.
Worked example: Kubernetes YAML Validator & Linter
A Deployment named Web_App with apiVersion: extensions/v1beta1, a selector matching app: api but a pod template labeled app: web, paired with a Service that defines two ports with no name field, produces exactly 4 errors and 4 warnings across those two manifests: the uppercase/underscore name fails the RFC 1123 check, the deprecated apiVersion is flagged (extensions/v1beta1 was removed from Kubernetes years ago — apps/v1 is current), the selector/label mismatch is caught explicitly (a Deployment like this would actually fail to ever match its own pods), and the Service's two unnamed ports trigger the "needs a name when there's more than one" rule. A correctly formed Deployment with matching labels, a pinned image tag, and both resource requests and a memory limit passes with 0 errors, 0 warnings.
Why selector/label mismatches matter so much
A Deployment's spec.selector.matchLabels isn't just documentation — it's the live query Kubernetes uses to decide which pods belong to that Deployment, and it's immutable after creation. If the selector doesn't match the pod template's own labels, the Deployment will either fail outright at apply time or, worse, silently fail to adopt the pods it just created, leaving you with pods running but a Deployment controller that thinks it manages zero replicas. This is exactly the kind of mistake that's easy to make by hand-editing YAML and easy to miss in review, which is why it's treated as a hard error here rather than a warning.
Why this isn't a replacement for kubectl --dry-run=server
This linter runs entirely offline, checking YAML syntax and a fixed set of best-practice and correctness rules against the manifest text alone — it has no concept of your actual cluster's installed CRDs, admission webhooks, resource quotas, or exactly which API versions your specific cluster version supports beyond the common ones hardcoded here. kubectl --dry-run=server sends the manifest to your real API server, which validates it against the live OpenAPI schema and any admission controllers you've configured — catching things like "this CRD doesn't exist" or "this namespace has a quota that would reject this request" that a static, offline check fundamentally cannot know about. Use this for a fast pre-commit sanity check; use the real dry-run before anything goes to a cluster that matters.
An empty box stays empty, and the success line pluralizes
Strict mode, which checks probes and runAsNonRoot, starts off. An empty box returns nothing, not the success sentence. When the manifests pass, the line reads that one manifest looks valid, or that several manifests look valid once the count is above one. Each problem is marked with a cross and ERROR, or an exclamation and warn, then where it sits and the message. The counts for manifests, errors, and warnings sit beside that text. I turn strict on only when I want those extra checks, because the default leaves them out.
Related tools
- YAML ↔ JSON Converter
Convert YAML to JSON and JSON to YAML for Kubernetes, Docker Compose, GitHub Actions and other config files.
- Online HTML, CSS & JS Editor (Live Preview)
Write HTML, CSS and JavaScript and see the result instantly — live preview, console, templates, shareable links and one-click HTML download. Like CodePen, with nothing to sign up for.
- JavaScript & TypeScript Playground
Run JavaScript or TypeScript in your browser with a real console — console.log, console.table, async/await, fetch and timers. Stop runaway loops, and share snippets as a link.
- Python Online Compiler & Playground
Write and run Python 3 in your browser — no install, no account. Real CPython (Pyodide) with input(), NumPy, pandas and Matplotlib charts, and shareable links.
- SQL Playground (SQLite Online)
Practice and run SQL in your browser on a real SQLite database — sample shop data included, joins, CTEs and window functions, import CSV/Excel as tables, open .db files and export results.
- AI API Cost Calculator (LLM Token Pricing)
Estimate what an LLM-powered feature will cost — tokens per request, requests per month, input/output prices per million tokens, prompt caching and batch discounts. Works for any model: enter your provider's current prices.