Skip to content

content.md

Latest commit

 

History

History
66 lines (49 loc) · 4.45 KB

content.md

File metadata and controls

66 lines (49 loc) · 4.45 KB
Raw

Content pipeline

What's shipping

Each of the 10 indexable tools has a hand-authored guide in src/data/toolGuides.ts. A guide has:

  • overview — one paragraph, tool-specific, no marketing filler.
  • workedExamples — 2–4 real inputs + expected outputs + why-it-matters explanation.
  • whenToUse — 3–4 concrete situations.
  • whenNotToUse — honest limitations. Do not skip this. Pages that only list positives read as promotional; pages that admit limits read as authoritative and get cited.
  • troubleshooting — symptom → fix pairs.
  • relatedSlugs — internal links to other real tools.
  • lastReviewed — ISO date.

The guide renders in two places:

  1. Server-side in the prerendered HTML (src/scripts/prerender.ts) — inside <noscript> and a hidden <div id="prerender-shell">. This is what Googlebot / OAI-SearchBot / PerplexityBot see before any JS runs.
  2. Client-side in ToolPageLayout (src/components/ToolGuide.tsx) — the actual visible version once React hydrates.

The two versions are identical content. Do not diverge them — Google flags cloaking.

Authorship disclosure

Guides are drafted with LLM assistance and reviewed by the XFree.in team before ship. The rendered page and JSON-LD credit "XFree.in team" as the author (an Organization, not a fake person). This complies with Google's stance on AI-assisted content: allowed if it's useful and honest about who's taking editorial responsibility.

Rule: if you haven't read a guide, don't ship it. Every TODO: string is a blocker.

Adding a guide for a new tool

  1. Wire the tool component in App.tsx and flip the tool's status to "indexable" in src/data/toolsRegistry.ts.
  2. Scaffold a stub: 
    npm run generate:guides -- --slug=your-new-tool-slug
    Copy the printed stub into TOOL_GUIDES in src/data/toolGuides.ts.
  3. Replace every TODO: with real content. Test each worked example against the actual tool — if the tool doesn't produce the output you claim, the guide is a lie.
  4. Rebuild: 
    npm run build:vercel
grep "Worked examples" dist/tools/your-new-tool-slug/index.html
    That grep must find the string in the raw HTML — that's your confirmation the guide is indexable pre-JS.
  5. Deploy. Ping IndexNow: 
    PUBLIC_SITE_URL=https://www.xfree.in npm run indexnow -- --url=https://www.xfree.in/tools/your-new-tool-slug

What you must NOT do

  • Do not use --gemini and ship what comes back unedited. The scaffolder deliberately does not auto-fill from Gemini for exactly this reason. Draft with Gemini in a separate session (interactive, so you can push back on hallucinations), then paste the reviewed result into the guide file.
  • Do not fabricate testimonials, reviews, or user comments. Not in guides, not on landing pages, nowhere. Real UGC is fine — solicit it, moderate it, attribute it. Fake UGC is illegal in most jurisdictions and detectable.
  • Do not clone the same guide across tools with a search-and-replace on the tool name. Google's scaled content abuse policy catches this. Every guide has to be actually different — different examples, different edge cases, different failure modes.
  • Do not add guides for tools whose React component doesn't exist or doesn't work. A guide promises the tool does X; if the tool doesn't do X, the guide is misleading. Ship the tool first.

Reviewing existing guides

For each guide already in TOOL_GUIDES, sanity-check:

  • Every worked-example input produces the claimed output when pasted into the actual tool.
  • Nothing in whenToUse overstates what the tool does.
  • whenNotToUse covers a real limit, not a fake "we're so honest here's a non-limitation" tell.
  • troubleshooting items are things you'd actually hit, not hypothetical.
  • relatedSlugs all exist in src/data/toolsRegistry.ts as indexable tools (the audit script catches this indirectly — a bad slug won't render, and the noindex lint doesn't cover it).
  • lastReviewed reflects the last time a human actually re-read the guide.

Bump lastReviewed when you re-verify, not automatically on every deploy. The date is a trust signal, not a marketing counter.