Build Verification - 44 Automated Checks on Every Deploy

Wire does not trust what it builds. After rendering every markdown page to HTML, it runs 44 automated checks across the finished site. Every check produces a specific, actionable message. Pages that fail do not ship silently. You see exactly what broke and where.

This is the final layer of Wire's quality system. The content quality layers work on markdown before and during writing. Build verification works on the finished HTML after rendering. It catches problems that only become visible in the final output: broken links between rendered pages, missing meta tags that templates should have inserted, images without dimensions, structured data with invalid JSON.

How It Works

After wire.build renders your site to HTML, run the verification:

python -m wire.lint --site /path/to/site

Wire parses every HTML file, builds a cross-page index (for duplicate detection and orphan checks), and runs all 44 rules. The output lists every failure with the rule ID, the affected URL, and what to fix.

You can also run specific checks:

python -m wire.lint --site /path/to/site --rules RULE-01,RULE-05,RULE-33

On a typical 200-page site, the full scan takes under 2 seconds. Zero API calls, zero cost.

What Wire Checks

The 44 rules fall into 11 categories. Each rule exists because independent research, not Google's public statements, shows it affects search performance or site health.

Page Structure (11 rules)

These checks verify that every page has the HTML elements search engines need to understand and rank it.

Canonicalization (4 rules)

Canonical tags tell search engines which version of a page is the "real" one. Getting this wrong splits your ranking signals across multiple URLs.

URL Structure (6 rules)

Clean URLs help search engines and users. These rules enforce the URL patterns that every major SEO study recommends.

Sitemap Integrity (4 rules)

Your sitemap tells Google which pages exist and matter. These rules catch contradictions between your sitemap and your actual pages.

Crawl Access (4 rules)

robots.txt controls which pages search engines can access. Mistakes here can accidentally hide your entire site.

Internal Links (3 rules)

Internal links distribute ranking power across your site. Broken links waste it. Orphan pages never receive it.

Security (1 rule)

Images (3 rules)

Multilingual (3 rules)

These rules only activate when your pages use hreflang tags for multilingual content. If your site is single-language, they are skipped automatically.

Structured Data (3 rules)

JSON-LD structured data helps Google understand your content type, author, dates, and relationships. Invalid structured data is worse than none at all.

Indexability (2 rules)

Where This Fits in Wire's Quality System

Wire enforces content quality at four levels. Each catches what the previous layers missed.

The first three layers work on markdown, the raw content. Build verification works on the finished HTML, what search engines and visitors actually see. A page can have perfect markdown and still fail verification if a template forgets to insert a canonical tag, or if a Jinja2 partial omits the viewport meta tag.

This is why Wire runs verification after rendering, not before. The template layer can introduce problems that do not exist in the content layer. Wire catches them before they reach production.

What Wire Does Not Check at Build Time

Six checks from the original specification require a live HTTP server and are deliberately excluded from build verification:

• Redirect chains (RULE-14): requires following HTTP redirects, which only work on a running server
• 4xx/5xx status codes (RULE-27): requires HTTP requests to external URLs
• External link validation (RULE-32): requires fetching external pages, which is slow and rate-limited
• SSL certificate validity (RULE-36): requires connecting to the live server
• Page load speed (RULE-38): requires a browser rendering engine (Lighthouse)
• Mobile rendering (RULE-39): requires device emulation

These are server-level and network-level checks. Wire is a build tool. It works with files on disk. For live-server checks, use Google Search Console's URL Inspection tool or a monitoring service like Pingdom or UptimeRobot.

Typical Results

On a real 316-page production site, the first build verification run found:

• 4 pages with duplicate H1 tags across different sections
• 12 internal links pointing to pages that had been renamed
• 2 pages where the template failed to insert Open Graph tags
• 1 page with a JSON-LD block containing a trailing comma (invalid JSON)

Total time: 1.8 seconds. Total cost: zero. Every issue included the exact URL and a specific fix instruction. The same issues would have taken a manual review hours to find, if they were found at all.

Running Selectively

Not every check matters for every site. A single-language site does not need hreflang rules. A site without structured data does not need JSON-LD rules. Run only what applies:

# Only check page structure and internal links
python -m wire.lint --site site/ --rules RULE-01,RULE-02,RULE-03,RULE-05,RULE-07,RULE-33,RULE-35

# Only check sitemap consistency
python -m wire.lint --site site/ --rules RULE-23,RULE-24,RULE-25,RULE-26

Every rule ID is stable. You can save your preferred rule set and run the same checks on every build.