Blogpost Editor

Steps

1) Resolve the post(s)

• Discover likely post files (index.mdx, index.md, .md, .mdx).
• Use AskUserQuestion if multiple candidates.
• Read the post file.

2) Parse frontmatter and body

Extract:

• title, date, summary/description/excerpt, tags/categories
• body text

If YAML is invalid, note it as Critical and stop further checks only if the post can’t be reliably parsed; otherwise continue.

3) Post type & reader intent classification

Infer (do not ask unless ambiguous):

• Post type: {personal story, tutorial, opinion, travelogue, review, medical/health journey, other}
• Likely target reader: {general, peers, future patients, niche experts}
• Primary reader goal: {learn, decide, be reassured, replicate, be entertained}

Use this to tailor editorial recommendations (e.g., checklists/timelines for medical journeys).

4) Structural checks (PASS/FAIL)

Record:

• Word count >300 (exclude code blocks)
• Frontmatter complete (title/date/summary/tags)
• Summary length 10–200 chars
• Headings present for posts >800 words
• Conclusion present (explicit closing thought)
• Links: no empty/placeholder URLs

5) Narrative & structure review

Evaluate:

Opening hook

• Flag generic throat-clearing or delayed stakes.
• Suggest a stronger hook using existing material (don’t rewrite, propose options).

Logical flow / ordering

• Identify sections that are out of order or would read better swapped.
• Provide a suggested outline using the author’s existing headings.

Heading hierarchy & scannability

• Flag long blocks without headings.
• Recommend a “reader map” if >1200 words:
  • 4–6 bullets summarizing what’s ahead.

Conclusion

• Check that the ending lands a takeaway or “what now”.
• Suggest a closing paragraph structure if needed.

6) Repetition & redundancy detection

Flag:

• repeated claims stated multiple times across the post,
• repeated phrases/frames (e.g., “delay as long as possible” repeated in multiple sections),
• back-to-back paragraphs with the same rhetorical move (“I then…” x3+).

For each:

• quote both instances,
• recommend “merge / keep one / shorten later reference”.

7) Factual risk & hedging review (non-academic, high value)

Identify “high-risk” factual claims:

• absolute statements: “no cure”, “only recourse”, “impossible”
• numerical claims: mortality %, durability years, risk reductions
• comparative pharmacology statements
• claims about evolving tech/procedures

For each:

• label: {needs citation, needs hedging, needs context}
• suggest a safer phrasing template:
  • “In my case / my team said…”
  • “My understanding is…”
  • “For predominant AR, options may be more limited than for calcific AS…”
• If you can’t verify within the post, recommend linking to guidelines or primary sources.

8) Fact-check: research contentious claims

From the claims flagged in step 7, select those that are contentious, surprising, or load-bearing (i.e., the post's argument collapses if the claim is wrong). Skip claims that are clearly personal anecdote ("my doctor told me…") or widely uncontested common knowledge.

For each selected claim:

1. Search — use WebSearch (2–3 targeted queries per claim) to find authoritative sources: peer-reviewed papers, official guidelines, reputable journalism, manufacturer specs.
2. Evaluate — read the top results with WebFetch and compare against the author's claim. Classify:
   • Wrong — evidence clearly contradicts the claim. Flag as Critical in the report. Quote the counter-evidence and provide source URLs.
   • Disputed — credible sources disagree or the claim is a known misconception. Cite the counter-evidence.
   • Partially supported — true with caveats (different population, outdated, context-dependent). State the caveat.
   • Supported — evidence agrees. Note the strongest source.
   • Unverifiable — no authoritative source found either way. Recommend the author cite their own source or soften.
3. Report — for each researched claim, output:
   • the claim (quoted),
   • verdict (wrong / disputed / partially supported / supported / unverifiable),
   • key source(s) with URLs,
   • suggested author action: keep as-is, add citation, add caveat, revise, or remove.

Guardrails for this step:

• Cap at 5 claims per post to keep the review actionable. Prioritize by reader-impact.
• Do NOT rewrite the claim for the author. Provide evidence and let them decide.
• When sources conflict, present both sides and flag the disagreement.
• Disclose when search results are inconclusive — never fabricate certainty.

9) Tone & sensitivity review (especially for medical journeys)

Flag:

• accidentally “salesy” praise of a doctor/hospital,
• overly absolute framing about other clinicians,
• fear-inducing lines that could mislead readers,
• humor that might land poorly given context.

Offer balanced alternatives that preserve voice (suggestions, not rewrites).

10) Reader usefulness module (genre completeness)

If post type is medical/health journey, check presence of (PASS/FAIL + suggestions):

• “Where I am now / current status”
• Simple timeline (table or bullets)
• “What I wish I knew”
• “Questions I asked / would ask”
• Practical checklist (what to pack / what helped)
• “Not medical advice” disclaimer (optional but recommended)

Output missing items as Recommended with specific insertion points.

11) Visual pacing & asset plan (editorial)

Beyond “add images”, propose:

• What visual to add (timeline table, diagram, pull-quote, callout)
• Where to place it (after which paragraph/heading)
• Why (what it clarifies)
• Optional: a draft caption + draft alt text (short)

Flag any wall-of-text stretch >300 words without a visual break and propose one visual break.

12) Spelling, grammar, and typos (NO auto-correct)

Scan for:

• typos, grammar issues, tense inconsistencies, ambiguous pronouns
• but respect wordplay and casual tone

For each flagged item:

• include the full sentence,
• suggest correction (do not apply automatically),
• label “possibly intentional” when uncertain.

13) Produce the editorial report

Format exactly:

Editorial Review: "{title}" Word count: {count} {PASS/FAIL} Reading time: ~{N} min (assume 220 wpm) Readability: ~grade {N} (Flesch-Kincaid, exclude code blocks) Frontmatter: {PASS/FAIL} Thumbnail: {PASS/FAIL} ({filename} or "missing")

Structural Checks Check Status Note ... Critical (must fix before publish) {Issue} — {location}

Quote: “...”

Why it matters:

Suggested fix:

Recommended (should fix) ...

Suggestions (nice to have) ...

Repetition & Redundancy Instance A: {location} — “...”

Instance B: {location} — “...”

Recommendation: ...

Factual Risk & Hedging Claim: "..."

Risk: {needs citation/hedging/context}

Suggested phrasing template:

Fact-Check Results Claim: "..."

Verdict: {wrong / disputed / partially supported / supported / unverifiable}

Evidence: ...

Sources: source title, ...

Action: {keep as-is / add citation / add caveat / revise / remove}

Visual Pacing & Asset Plan Wall-of-text: {heading} (~{N} words)

Add: {timeline/diagram/callout/pull-quote}

Placement:

Caption/alt (optional):

Tone Overall assessment:

Flags:

Summary Critical: {count}

Recommended: {count}

Suggestions: {count}

Verdict: {Ready to publish / Needs revision / Major rework needed}

---37:["$","$L3f",null,{"content":"$40","frontMatter":{"name":"blogpost-editor","description":"Deep editorial review for publication readiness — narrative, structure, scannability, factual risk, tone, and reader usefulness. No silent rewrites."}}]