Introduction
If you've shipped anything to the web in the last decade, you've probably bumped into the Front-End Checklist. It's the GitHub repo with 73k+ stars that engineers quietly reference the night before a launch to make sure they didn't forget rel="noopener" or an og:image. In 2026, Front-End Checklist grew up: it now has a web app in beta and — more interesting to me — an MCP server that lets Claude and other AI agents run the checklist against your code directly.
I've been using it for pre-launch audits and as a guardrail inside AI coding sessions for the last few weeks. This is what it actually is, what it isn't, and whether it deserves a slot in your workflow.
Key Features
The rule corpus (the real product)
The core is still what it always was: 350+ rules spread across 11 categories — HTML, CSS, JavaScript, images, fonts, accessibility, performance, security, SEO, PWA, and a launch/pre-flight bucket. It's opinionated but sane. Nothing exotic, nothing cargo-culted from a 2018 blog post that everyone kept copy-pasting. Each rule has a short justification and, where it matters, a link to the spec or the MDN page.
Curated checklists
Instead of dumping all 350 rules on you, Front-End Checklist ships pre-built checklists for common scenarios: pre-launch audit, junior dev PR review, AI-assisted quality check. This is the part that surprised me — decision fatigue is real, and being handed a 40-item "about to ship" list beats staring at the full rule catalog trying to figure out what applies today.
MCP server integration
This is the 2026 upgrade that matters. The MCP server exposes the rule corpus as tools an AI agent can call. In practice, that means I can tell Claude "run the pre-launch checklist against this repo" and it walks through the rules, inspects files, and reports back with a structured audit. It's not magic — the agent still has to actually read your code — but the rules and rationale come from a curated source instead of whatever the model half-remembers from training.
Progress tracking
The web app tracks which rules you've marked done per checklist run. Useful for a single sitting. Not useful across sessions, which I'll come back to.
Pricing Breakdown
There's one plan. It's free.
| Plan | Price | What you get |
|---|---|---|
| Free | $0 | All 350+ rules, all curated checklists, MCP server, GitHub source |
No paid tier, no "pro" features gated behind a login, no enterprise upsell. The project is open source and appears to be maintained as a genuine public good. If that changes, I'll update this. For now: zero cost, zero lock-in.
Pros
- Actually free and open source. 73k+ stars on GitHub, MIT-adjacent licensing, forkable. You are not going to wake up to a pricing email.
- MCP integration is the killer feature in 2026. Plugging a curated rule set into an AI coding agent closes the loop between "I know I should check X" and "the agent just checked X for me."
- Breadth in one place. Accessibility, perf, security, SEO — you'd normally stitch this together from Lighthouse, axe, and three blog posts. Here it's one corpus.
- Curated checklists cut decision fatigue. The pre-launch and PR-review lists are genuinely well-scoped.
Cons
- The web app is in beta and it shows. Occasional layout glitches, and I hit one state where marking rules complete didn't persist on refresh. Not blocking, but not polished.
- No cross-session sync or team collab. If you close the tab, your progress is gone. For a solo audit right before deploy that's fine; for anything team-shaped, it's not the tool.
- Rules are opinionated defaults with limited customization. You can skip rules, but you can't easily maintain your own overrides or a house style layered on top. If your org has a strong opinion that diverges, you'll fight it.
- No CI/CD hooks or automated scanning. This is a checklist, not a linter. Nothing runs on a PR unless you (or your agent) manually invoke it. If you want gated CI checks, pair it with Lighthouse CI, axe-core, or ESLint.
Who Is It For
Solo builders and small teams shipping to production. If you're the one person doing the pre-launch pass, this is exactly the tool. The curated checklists match how you actually think the night before a deploy.
Anyone running AI coding agents. The MCP server is the reason to pay attention this year. If Claude or a similar agent is already writing code in your repo, giving it a rule corpus to grade against is a real quality bump. It's not a substitute for review — it's a floor.
Not for: teams that need audit trails, multi-reviewer workflows, or CI-gated quality checks. There are heavier tools for that (and they cost real money). Also not for anyone who wants a fully custom rule DSL — the checklist is the checklist.
Verdict
Front-End Checklist earns a 7.5/10 from me. The rule corpus has been quietly excellent for years, and the MCP server upgrade in 2026 is the first thing that made me actively reach for it during coding sessions instead of just bookmarking it for launch day. It's free, it's open source, and the ceiling is honest — this is a curated reference, not a scanner or a platform.
Recommendation: install the MCP server today if you're running AI coding agents. It takes about ten minutes and immediately makes agent-written code answer to a real rule set. For pre-launch audits, keep it bookmarked and run the pre-launch checklist before every production deploy. If you need CI enforcement or team workflows, pair it with something automated — don't try to make this be that tool. But for what it is, at what it costs, it belongs in the stack.