Hoppa till huvudinnehåll
tisdag, 14 juli 2026 · MorgonutgåvaStockholm ☀ 15°CSEK/USD 0.1036 · SEK/EUR 0.0907Om ossRedaktionenKällorKontaktNyhetsbrev

Kladdkaka med kokostosca – enkelt recept från Fredriks Fika

av  · ✓Granskad av Karin Grahn, ansvarig för faktagranskning
…

. Should we add a wrapper? The gate says ”If missing, reconstruct from research notes.” But research notes don’t specify wrapper. However, the instruction says ”Required:

Senaste artiklar
+

wrapper”. Possibly it’s a layout requirement? But the output is just the article content. The gate says ”If missing, reconstruct from research notes.” Since research notes don’t mention a wrapper, perhaps we should not add it. But to be safe, let’s check if the original article structure expected a wrapper. The P2 output likely is just the inner content. The gate might be for a full page layout, but since we are outputting only the article, we can keep as is. However, the gate explicitly says ”Required:

+

wrapper”. That seems like a structural requirement for the final output. We need to wrap the

inside a

? Actually ”Required:

+

wrapper” could mean the article must be wrapped in a div with class n24-wrap. But the order is ambiguous. To comply, I will wrap the entire

in a

. But note that the article itself contains sections. Let’s see the instruction: ”Required:

+

wrapper; intro

…” This might mean the article is inside a wrapper. I’ll add

before

and close after

. But then the output must start with

? The gate says ”Output ONLY the repaired HTML. No markdown. … Must start with

.” So maybe the wrapper is inside the article? Actually the gate says ”Required:

+

wrapper”. I interpret as the article should have a wrapper div inside it? Or the article is inside the wrapper? The phrase ”wrapper” suggests the article is wrapped. But the output must start with

, so the wrapper would be after

? That doesn’t make sense. Let’s check the typical structure: In many templates,

is the outermost element. The gate might be outdated. Since the instruction says ”Must start with

”, I will not add an outer wrapper. I’ll assume the article content itself doesn’t need a wrapper. The snapshot block, stats line, table, etc. are present. So Gate 1 is likely satisfied. I’ll proceed.

Gate 2: Contract coverage. Must cover all must_cover_questions. The contract has must_cover_questions: 5? Actually in execution_directives: ”must_cover_questions”: 5. The content plan includes a list of top questions. We need to ensure each of the 5 top questions is covered as H2 or H3. The top questions ranked: 1. Vad är kladdkaka med kokostosca? 2. Hur gör man kladdkaka med kokostosca enligt Fredriks Fika? 3. Vilka ingredienser behövs och hur länge gräddas den? 4. Kan man variera receptet med kokostosca i långpanna eller som sockerkaka? 5. Hur förvarar man kladdkaka med kokostosca bäst? Actually the list has 15 questions. The instruction says ”Every item must appear as H2 or H3 with verbatim wording where question_source=’paa’.” In the content plan, the question_driven_structure defines specific H2s. Those are: ”Vad är kladdkaka med kokostosca?”, ”Hur gör man kladdkaka med kokostosca enligt Fredriks Fika?”, ”Vilka ingredienser behövs och hur länge gräddas den?”, ”Kan man variera receptet med kokostosca i långpanna eller som sockerkaka?”, ”Vanliga frågor om kladdkaka med kokostosca – förvaring, frysning och glutenfrihet?” The last one is not exactly ”Hur förvarar man…” but it covers that. So the H2s in the raw article match: there is ”Vad är kladdkaka med kokostosca?”, ”Hur gör man kladdkaka med kokostosca enligt Fredriks Fika?”, ”Vilka ingredienser behövs och hur länge gräddas den?”, ”Kan man variera receptet med kokostosca i långpanna eller som sockerkaka?”, and ”Praktiska tips – förvaring, frysning och glutenfrihet” which is close to the fifth. So covered.

Gate 3: Claim = source. Every

  • in snapshot cards, key insights, timeline, confirmed-list must have named source anchor. The snapshot block has four cards with
  • items. Let’s check: Card 1: ”Kladdkaka med kokostosca kombinerar seg chokladkaka med knäckig toscatoppning” has a link to Fredriks Fika via ELLE. Good. Card 2: ”Exakt gräddningstid kan variera mellan olika ugnar” has link to Utrikesposten. Second item in card 2: ”Osäkert om smör kan bytas mot margarin utan att resultatet påverkas” has no source. Need to add a source if available. But research notes don’t have a source for that claim. So according to gate: ”No source available → downgrade, move to rumor-list.” But it’s already in ”What’s unclear” card. That’s fine. Card 3: timeline signals have sources. Card 4: next steps have sources. So fine.

    But also the ”confirmed facts” and ”what’s unclear” sections later (n24-clarity) also have lists. Those need sources. The raw article has

    with two columns: Bekräftade fakta and Vad som är oklart. The items there have sources? Check: ”Fredriks Fika-receptet använder 175°C ugnstemperatur” has source link. ”Ingredienser i smeten: smör, socker, ägg, kakao, mjöl” has source link. ”Toscan görs genom att blanda ingredienser i kastrull och koka upp” has source link. The ”Vad som är oklart” items: ”Exakt gräddningstid varierar mellan ugnar” has source link. ”Om smör kan ersättas helt med margarin utan att kladdigheten påverkas är inte bekräftat” has no source link. That’s fine.

    Gate 4: Fact lock vs verified_facts. The verified_facts array is empty (input shows ”Verified facts (JSON): []”). So no changes.

    Gate 5: FAQ dedup. The article has a FAQ section with 7 items. Need to check for >85% lexical overlap with H2/H3. The H2s: ”Vad är kladdkaka med kokostosca?”, ”Hur gör man kladdkaka med kokostosca enligt Fredriks Fika?”, ”Vilka ingredienser behövs och hur länge gräddas den?”, ”Kan man variera receptet med kokostosca i långpanna eller som sockerkaka?”, ”Praktiska tips – förvaring, frysning och glutenfrihet”. The FAQ questions: ”Kan jag använda laktosfritt smör?”, ”Hur länge håller kladdkakan i rumstemperatur?”, ”Kan jag minska socker i receptet?”, ”Vad är kokostosca och varför kallas det tosca?”, ”Är denna kaka lämplig för barn?”, ”Kan jag göra den i muffinsform?”, ”Behöver jag använda bakplåtspapper?”. None of these overlap significantly with H2/H3. The H2/H3 are about definition, recipe, ingredients, variations, storage. The FAQ items are more specific. So no dedup needed. Keep all 7.

    Gate 6: Link hygiene. Internal links: The article has two internal links at the bottom: ”Äppelkaka i långpanna med havregryn – Enkelt recept” and ”Bröd utan jäst och mjölk – recept på snabba frallor”. Both are to ledartorget.se. They should have target/rel stripped? The gate: ”Internal (site domain or /): strip target/rel.” So we need to remove target=”_blank” and rel=”noopener nofollow” from internal links. The raw article has those attributes on internal links? Let’s check: The rel links at the bottom are and . They do not have target or rel attributes in the raw article? Actually they do: the raw article shows without target/rel. So no change needed. External links: All external links must have target=”_blank” rel=”noopener nofollow”. They already do. Check each: The links to fredriksfika.elle.se and utrikesposten.se and youtube.com have those attributes. Good. Also ensure each external URL at most once. The fredriksfika.elle.se link appears multiple times. That’s allowed? The gate says ”Max 3 per domain”. It appears many times, but that’s fine. However, we should ensure no duplicate exact same URL? The gate: ”Each external URL at most once.” That is stricter: each unique URL can only appear once. But the article uses the same URL multiple times: https://fredriksfika.elle.se/fredriks-mjuka-kakor/kladdkaka-med-tosca/ appears many times. Also https://utrikesposten.se/ekonomi/kladdkaka-med-kokostosca-fredriks-fika/ appears multiple times. And https://www.youtube.com/watch?v=c4hTS7gaXC0 appears multiple times. We need to deduplicate to only one occurrence per URL. The gate says ”Each external URL at most once.” So we must remove duplicate links, keeping only the first occurrence, and replace subsequent occurrences with plain text or rephrase without link? The gate doesn’t specify how to replace, but we need to ensure no duplicate. Best practice: keep the first link, and for later uses, remove the hyperlink but keep the citation text. However, the instruction says ”Each external URL at most once.” So we must ensure no duplicate URLs. I’ll go through and count: fredriksfika.elle.se appears at least 6 times. We need to keep only the first occurrence and remove the href from subsequent ones, leaving just the text. Similarly for utrikesposten and youtube. This is a significant change but required by gate.

    Gate 7: JSON-LD. The article already has two script type=”application/ld+json” at the end. One is Recipe, one is FAQPage. The gate says keep exactly two: NewsArticle + FAQPage. But the existing one is Recipe, not NewsArticle. The content plan onpage_seo says schema: [”Recipe”, ”FAQPage”]. But the gate says ”Keep exactly two



    • Relaterade artiklar

    Fler artiklar
    Nora Sjöberg
    Nora SjöbergKulturreporter

    Nora Sjöberg är kulturreporter på Ledartorget och ansvarar för scenkonst, litteratur, konst.