25%
Conversion
Can a stranger understand the offer and take one action?
- Email capture and a specific CTA
- Social proof and risk reversal
- Low friction, human contact
Open source · Apache-2.0 · zero dependencies
Grade your launch
before the internet does.
LaunchGrade scans any public page in about a second. 54 checks across conversion, trust, technical, SEO, social, accessibility and AI-agent readiness — then it hands you the fixes that matter, in priority order.
npx -p github:fastpresspages/launchgrade launchgrade your.site
No account. No tracking. No paid API. Node 18.17+ and one command.
- 53checks
- 7categories
- 0runtime deps
- <1stypical scan
terminal report◆
standalone HTML◆
markdown◆
json◆
svg badge◆
fix plan◆
MCP server
The 60-second pre-launch check
You ship. It checks.
You fix the right thing.
No dashboard to learn, no browser to install, no score without a next step. Point it at a URL and read.
-
01
Scan the public page
Fetches the HTML, headers and a small allowlist of well-known files: robots, sitemap, llms.txt, security.txt, AGENTS.md. No JavaScript execution, no cookies.
launchgrade your.site -
02
Get a weighted grade
Every check returns pass, warn, fail or info. Categories are normalized and weighted, so conversion and technical issues outrank cosmetic ones.
A+ · 100/100 · launch-ready -
03
Fix the top three
The fix plan sorts by impact, not by checklist order. Each item explains the failure in plain language and tells you what to change.
launchgrade your.site --out reports
Want a second pair of eyes? The free audit is a human, not a bot, reading your page.
Ask for a free auditWhat it actually looks at
Seven categories.
Fifty-three honest checks.
Each category carries a weight. Nothing checks for your favorite framework, and nothing rewards fake urgency or keyword stuffing.
15%
Trust
Would a careful buyer believe you and feel safe?
- Privacy, terms and about links
- Real social presence
- security.txt and current copyright
20%
Technical
Does the page load fast, securely and indexably?
- HTTPS, status, speed, compression
- Viewport, canonical, language
- Mixed content and analytics
15%
SEO
Can search engines understand and rank the page?
- Title, description and headings
- Sitemap, robots and schema
- Image alt text and text ratio
10%
Social sharing
Will a shared link look credible in the feed?
- Complete Open Graph card
- Twitter/X summary card
- Share image and alt text
5%
Accessibility
Can everyone use the page, including keyboard and screen-reader users?
- Form labels and button names
- Descriptive links and landmarks
- Pinch zoom and skip link
10%
AI-agent readiness
Can ChatGPT, Claude, Perplexity and coding agents read and recommend you?
- llms.txt and explicit crawler access
- AGENTS.md and Markdown negotiation
- FAQ, semantic HTML, no-JS copy
Bring your own check.
A check is a few lines of readable JavaScript and a fixture. If it would have saved your launch, it belongs here.
Read the contributor guideOne scan, five artifacts
Reports you can
actually use.
Terminal output for the moment you ship. A standalone HTML report for the team. JSON for your dashboard. Markdown for the issue. A badge for the README. And a fix plan that reads like a senior friend looked at the page.
launchgrade · sample
$ npx -p github:fastpresspages/launchgrade launchgrade your.site
▲ LaunchGrade v0.3.0 · your.site
┌──────────────────────────────────────────────┐
│ A+ 100/100 Launch-ready │
└──────────────────────────────────────────────┘
Conversion 100 ████████████████████
Trust 100 ████████████████████
Technical 98 ███████████████████░
SEO 100 ████████████████████
Social sharing 100 ████████████████████
Accessibility 100 ████████████████████
AI agents 100 ████████████████████
Fix these first
1. WARN No analytics (deliberate on this site)Send it to the team
A single dark file with the grade ring, every category and all 54 checks. No scripts, no external assets.
Turn a score into work
Ordered by impact, with the failure, the reason and the concrete change. Paste it into an issue.
Show your grade
A plain SVG badge for your README. Update it on every release with one action step.
Built for humans and agents
Your coding agent
can run the audit.
LaunchGrade ships a zero-dependency MCP server and a portable skill. Ask Claude, Cursor or any MCP client to grade a page, explain the failures and write the fixes — using the same checklist a human would.
- grade_launchScan a URL and return Markdown or JSON.
- get_fix_planRanked fixes, ready to paste into a task.
- list_checksEvery check, category and weight.
claude_desktop_config.jsonmcpServers
{
"mcpServers": {
"launchgrade": {
"command": "npx",
"args": [
"-y", "-p",
"github:fastpresspages/launchgrade",
"launchgrade-mcp"
]
}
}
}
Then ask:
“Grade example.com. Fix the three highest-impact issues. Do not invent metrics.”
Free, human, no pitch
Want a second
pair of eyes?
Run the CLI first. If you want a human read, send the URL and what you are launching. You will get a short, prioritized reply — not a 40-page PDF, not a calendar link, not a drip sequence.
- The three fixes that would change the most
- What is already working, so you do not break it
- A straight answer on whether the page is ready
Also open to: collabs, sponsorships, tool partnerships and honest distribution deals. Say what you can offer and what you need. No paid links or fake reviews.
free
Request an audit
One human reply. No account, no payment.
Built in public
A small lab for
people who ship.
FastPress Labs is the community behind LaunchGrade and YouLauncher. The plan is simple: make one useful thing at a time, document it honestly, and help the next person launch with fewer avoidable mistakes.
53checks shipped
7categories
0runtime dependencies
1real sample audit
Use the tool
Run it on your page. Ignore the checks you disagree with. Tell us which one was wrong.
GitHubBring a check
Small, specific, testable. If it would have saved your launch, it is probably a good check.
Open an issueWork together
Collabs, sponsorships, distribution or a genuinely useful tool. Tell us what you can offer.
Email usStraight answers
Questions,
answered plainly.
Still stuck? Send the form or email meetcode99@gmail.com.
Is LaunchGrade really free?
Yes. The CLI, GitHub Action, MCP server, reports and agent skill are open source under Apache-2.0. The human audit request is free too. There is no paid tier, no account and no tracking.
Does this replace Lighthouse?
No. Lighthouse measures browser performance and best practices. LaunchGrade checks the launch surface: offer clarity, capture, trust, share cards, search basics, accessibility signals and AI-agent readiness. They complement each other.
Do you store the URLs I scan?
The CLI runs locally and sends nothing to us. If you submit the audit form, the URL and message are delivered to the FastPress Labs inbox through FormSubmit so a human can reply.
Can I use it in CI?
Yes. Use the composite GitHub Action or run the CLI with --fail-under 80. The job exits with code 2 below your threshold, so a launch page cannot silently regress.
Does it work on private pages?
No. LaunchGrade fetches public URLs without your cookies or session. For a private page, export the HTML and use the library API with buildContextFromHtml, or ask for a human audit.
Why is it called LaunchGrade?
Because “launch” is the moment everything you skipped becomes visible, and a grade is easier to act on than a 60-item checklist. The letter is a summary; the fix plan is the product.
Who is behind it?
FastPress Labs, a small open-source community for people who ship. YouLauncher is its first trial project. LaunchGrade is the first shared tool. No venture funding, no growth team, no fake logos.
One command. No signup.
Find out what your
launch is missing.
npx -p github:fastpresspages/launchgrade launchgrade your.site
If it catches one broken share card before launch, it paid for itself. It is free, so that is easy.