A development preview can look complete while the production export is missing a route, an image, or a working contact link. For a small-business site, verification should target the static files that will actually be published.

This documented guide uses a real-estate company as an example. Property listings remain in the existing listing system; the static website explains the company, service area, process, and contact route.

Define the static boundary first

I operate a small real-estate business. Separate the approved company information from live property listings, customer records, appointments, applications, contracts, and payments. Mark missing facts as UNKNOWN. Do not invent listings, prices, transaction results, licenses, reviews, service areas, or guarantees. Propose the smallest public website and identify every function that must remain in an external managed service.

Convert a visual reference into rules

Use the attached reference image only for visual direction. Do not copy company names, text, numbers, testimonials, or claims from it. Create a restrained property-business design rule sheet with a clear grid, readable service-area information, factual process sections, optimized images, and one approved inquiry action. Wait for approval before implementation.

Require a static production build

Build the approved website with the Next.js App Router and output: 'export'. Keep live property listings, search, accounts, applications, contracts, payments, and customer records in external systems. Do not use a database, Next.js API Routes, Server Actions, request-time rendering, or private environment variables. Before coding, list any requirement that cannot be exported as static files.

A useful verification pass checks the artifact, not only the source:

1. Run the production build.
2. Confirm that the configured output directory exists.
3. Open every generated HTML route directly.
4. Check that local assets resolve from the exported path.
5. Search the output for UNKNOWN, placeholder text, private paths, and secrets.
6. Test external inquiry and listing links.
7. Check a narrow mobile viewport and keyboard navigation.

Do not call the site verified when only the development server works.

Install Sakupa MCP and fully restart

This publishing example uses Sakupa MCP, which is developed by the author.

{"mcpServers":{"sakupa":{"command":"npx","args":["-y","@sakupa/mcp@latest"]}}}

Save the configuration, then fully quit and restart the AI tool. MCP will not reliably appear in the existing process. Reopen the project and verify the tools before any side effect.

Confirm that Sakupa MCP is connected and that the init, analyze, and deploy tools are available. Do not initialize, analyze, upload, or publish anything yet. If unavailable, verify the configuration and restart the AI tool again.

Inspect the exact artifact that will be public

Use Sakupa to prepare the current project for first use. Do not publish anything.

Use Sakupa to inspect the current project and identify the real static output. Inspect only; do not upload. Report the selected directory, generated routes, public assets, files that must stay private, and unresolved checks.

The inspection should select the build output rather than the repository root. Stop if it includes .env, credentials, unpublished property data, customer files, or source documents.

Keep publication behind confirmation

Use Sakupa to publish the website that passed inspection. Before the first public action, explain who can access it, how long the free website remains available, and which management credential must stay private. Wait for my explicit confirmation before publishing. Return the public URL and expiration time with timezone.

The current manual states that a free site is public for 24 hours. Keep .sakupa/site.json private and out of public repositories.

After publication, open the returned URL in a signed-out browser and repeat the route, asset, contact-link, and mobile checks against the remote page. Record the checked URL and time instead of assuming that a successful command proves continued availability.

References

• Next.js: Static Exports
• Sakupa Website-building Guide
• Sakupa MCP deployment manual

Generative AI was used for structure and editing. Product instructions were reviewed on August 3, 2026; no end-to-end benchmark is claimed.