Article-writing prompt — read this first

This page is the canonical source of truth for every article published on RTI Wiki.
Both Claude and ChatGPT Codex must read it before writing any new content. Update here, not in chat.

Quick answer

When you generate an article for RTI Wiki, you are not writing “an article about X” — you are building the destination citizen-first answer in India for that query. If a smarter article exists elsewhere on the open web, we have failed.

Inputs you may be given

Query / topic
GSC metrics from the last 28 days: impressions, clicks, CTR, average position
An existing slug to refresh, or “new” for a brand-new article

Step 1 — Classify the query in 3 lines

Bucket	Trigger	Strategy
High opportunity	high impressions + low CTR + position 5-20	better title + meta + intro hook
Quick win	position 8-15 + already getting clicks	add depth — schema, FAQ, internal links
Content gap	relevant query but no dedicated page	brand-new ultimate-guide article
Topical authority	pillar / cluster expansion	cross-link siblings, become the hub

State which bucket and why in one sentence.

Step 2 — Mandatory article structure

Every article must follow this skeleton:

Frontmatter — htmlmetatags block (≤165 char description, comma-separated keywords)
Title — H1, ≤70 chars, ends with “— citizen guide 2026”
<WRAP info> quick answer — 40 to 60 words, direct answer to the query
What X is — in 50 words — exact 40 to 60 words for feature snippet
Legal position in India — Act + section, framework, authority structure, RTI relevance, case-law precedent
Step-by-step process — numbered DokuWiki list
Documents required — bullet list
Common mistakes — bullet list, each with a §-citation
Real-life example — <WRAP center round box> with name, district, dates, exact figures, total cost

Sample RTI letter (if applicable) — `

` block with §6(1), §10, §6(3), §7(1), §19(1) citations
  - **FAQ** — 6 to 10 questions, each as `==== Q ====` H4
  - **Sources** — bullet list of authoritative URLs
  - **Related on RTI Wiki** — 5-8 internal cross-links

===== Step 3 — Title + meta: 3 variants, pick strongest =====

Generate three. Each must satisfy the **HARD social-readiness rules**:

  * Title ≤ 70 chars
  * Meta description ≤ 165 chars
  * First paragraph ≥ 100 chars and standalone (works as a tweet)

Variant brief:
  - **Curiosity** — opens a loop ("Why does X take 14 months — and how to unstick in 31 days")
  - **Urgency** — frames cost of inaction ("Stuck in mutation? File this RTI before day 90")
  - **Clarity** — most scannable ("Sample RTI for land mutation delays — citizen guide 2026")

Pick the strongest, ship it. List the others below the article for git history.

===== Step 4 — Internal linking =====

Every article links to **at least 3** of these primary anchors:

  * [[https://righttoinformation.wiki/tools/ai-rti-draft-app.html|AI RTI Drafter]]
  * [[https://righttoinformation.wiki/tools/first-appeal-app.html|First Appeal Builder]]
  * [[https://righttoinformation.wiki/tools/awaaz-rti.html|AwaazRTI (voice)]]
  * [[https://righttoinformation.wiki/tools/pio-reply-checker-app.html|PIO Reply Checker]]
  * [[https://righttoinformation.wiki/tools/timeline-calculator-app.html|Timeline Tracker]]
  * [[:act|RTI Act, 2005]]
  * [[:cases|Case-law database]]
  * [[:rules|State RTI rules]]
  * [[:rti-fees-by-state|State-wise RTI fees]]
  * [[:guide:applicant:application:sample:start|Sample RTI letters index]]

**NEVER** use ''[[:tools:slug|...]]'' — those are static HTML, DokuWiki marks them broken. Always full https URL for /tools/ and /intelligence/ files.

===== Step 5 — Trust signals (3 minimum per article) =====

  * **Statute** — Act + section + year (e.g., "RTI Act 2005, §6(1)")
  * **Government body** — official ministry / commission / portal with a hyperlink
  * **Case law** — Supreme Court / High Court / CIC ruling with citation `(YYYY) X SCC Y`

Cite in-line, not as footnotes. Citations make the article quotable.

===== Step 6 — Schema markup (JSON-LD) =====

Every article embeds:

<code>
<html>
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@graph": [
    {"@type": "Article", "headline": "…", "datePublished": "…", "dateModified": "…", …},
    {"@type": "FAQPage", "mainEntity": [{"@type": "Question", …}, …]},
    {"@type": "HowTo", "name": "…", "step": [{"@type": "HowToStep", …}, …]}
  ]
}
</script>
</html>

ONLY <script type=“application/ld+json”> is allowed inside <html> blocks. DokuWiki escapes inline <style>, <svg>, and multi-element <div> blocks as raw text.

Step 7 — Visuals

Plan ONE infographic + ONE feature image per article:

Infographic — DokuWiki-native: <WRAP center round info 100%> + 4-column metric table + emoji-numbered process flow (① → ② → ③ → ④ → ⑤). Apple-liquid-glass green palette.
Feature image — describe in commit, generate separately, save to /home/bighelpers/wiki/data/media/social/auto/{slug}.png at 1200×630. ONLY embed if it actually exists; broken auto references render as red wikilink2 links.
SVG that needs custom CSS — render to a standalone .svg file under data/media/, embed via .

Generate three variants:

Twitter — under 260 chars (incl. URL), strong factual hook, 1 hashtag max
Facebook — 80-120 words, plain English, ends with URL
Long form — 4-6 short paragraphs: problem → law → process → CTA

The Buffer automation picks these up via the OG meta tags + AI rewriting.

Step 9 — CTR opening hooks

Every article opens with ONE of:

A specific number — “₹4.7 lakh recovered in 41 days”
A named protagonist — “Manjit Kaur exposed a phantom toilet”
A statutory clock — “30 days under §7(1) — silence is deemed refusal”
A direct question — “Mutation stuck after 14 months? Here is the §149 RTI.”

Avoid generic openers like “In India, the RTI Act allows…”.

Step 10 — Update strategy

Day 0 — IndexNow ping
Day 7 — re-check GSC; if impressions present but CTR <1%, regenerate title + meta
Day 30 — refresh: add new FAQ entries, update real-life example dates, re-verify statutory citations
Quarterly — link audit (we have a recurring job)

Hard rules (these are HARD failures, not warnings)

Title >70 chars → fails social audit
Meta description >165 chars → fails social audit
First paragraph <100 chars OR hidden inside <WRAP> → fails social audit
slug wikilinks → broken (use full https URL)
Inline <HTML> blocks with <style> or multi-element <svg> → DokuWiki escapes them as <pre> text
Image references that 404 → red wikilink2 link
No statute / case-law citations → article reads like opinion
No real-life example → article is forgettable
No internal links to RTI Wiki tools → user dead-ends, no funnel
Smart quotes inside HTML attributes → DokuWiki kills the parsed HTML

Output format (when you produce an article)

Return a single block in this exact order:

Keyword analysis (3 lines max)
Title + meta — 3 variants (with char counts)
Chosen title + meta (with rationale)
Article body — full DokuWiki source, ready to paste
Internal links used (5+ list)
Schema JSON-LD (separate code block)
Social copy (Twitter / FB / long-form)
Image + infographic plan (1-2 sentences each)
Update strategy (specific dates)

Do not produce explainers about what you are doing — just deliver the artefact.

Special instructions for India

Indian audience first — every example, scheme, helpline, address is Indian
Plain English, short sentences, no jargon, no Latin except statutory references
All currency in ₹ (not “Rs” except inside fixed RTI clauses)
Helplines, portals, departments must be real and verified
If a state rule changed after the latest elections, check the state gazette before publishing
Cite the 2023+ MPLADS Guidelines, 2022+ Adoption Regulations, 2026+ scheme amendments — never an outdated version

Where this file lives

Public URL (this page): https://righttoinformation.wiki/about/article-writing-prompt
Server canonical: /home/bighelpers/wiki/data/pages/about/article-writing-prompt.txt
Static markdown copy (for AIs that prefer plain markdown): https://righttoinformation.wiki/article-writing-prompt.md
Claude memory pointer: local memory file reference_article_writing_prompt.md
ChatGPT Codex pointer: in your system prompt, paste this URL and instruct: “Read https://righttoinformation.wiki/about/article-writing-prompt before writing any article. Apply every rule. Treat conflicts in chat as overridden by this file.”

When this page changes, both AI assistants pick up the new rules on next run because they fetch the URL fresh.

Last revised: 4 May 2026. Supersedes the v1 prompt and the earlier 10-step SEO engine.

article-writing-prompt, content-engine, seo, style-guide, rti-wiki, claude, chatgpt-codex, 2026