We need to output JSON with fields: joke_text, headline, meme_caption, image_pro...

We need to output JSON with fields: joke_text, headline, meme_caption, image_prompt, meta_title, meta_description. Must be in German, no markdown code block. Must be unique, start with Chuck Norris or mention him. Must be classic Chuck Norris style. Must avoid repeats. Provide a joke text in German, starting with "Chuck Norris". Provide headline max 80 chars, meme_caption max 80 chars, image_prompt English description for image generation, meta_title max 60 chars, meta_description max 160 chars. We need to ensure uniqueness and not repeat existing jokes. Let's craft a new joke: Something like "Chuck Norris can divide by zero." But that's common. Need something unique. Maybe: "Chuck Norris once tried to teach a cat to bark, and the cat now runs a successful dog show." That's weird. Better: "Chuck Norris can make a round square." That's also common. We need something more creative: "Chuck Norris once challenged a calendar to a fight, and now every month ends on a Friday." That's okay. But must be typical Chuck Norris genre: exaggeration, impossible feats, mythical powers. Maybe: "Chuck Norris can fix a broken heart with a single roundhouse kick, and the heart then becomes a black hole." That's too weird. Let's craft: "Chuck Norris once tried to count to infinity, but the numbers started to fear him and stopped at 42." That's referencing Hitchhiker's Guide; maybe okay. But need to be unique. Let's think of a scenario: "Chuck Norris can make a snowstorm in the Sahara just by blinking." That's good. Make sure it's unique and not a known joke. We'll produce: joke_text: "Chuck Norris kann einen Schneesturm in der Sahara erzeugen, indem er nur einmal blinzelt – und plötzlich fällt der erste Schnee, weil die Erde Angst hat, ihn zu ignorieren." But must start with Chuck Norris. So "Chuck Norris kann..." That's fine. Now headline: maybe "Der Schneesturm im Wüstenparadies". Must be max 80 chars. Meme caption: "Wenn Chuck Norris blinzelt, wird die Wüste zum Winterwald". Must be max 80 chars. Image prompt: English description: "A hyper-realistic portrait of a tough, legendary action hero man in his 50s with reddish-brown hair, thick reddish beard, piercing eyes, muscular build. Black sleeveless vest. Chuck Norris blinking, causing a snowstorm to erupt over a desert landscape, sand turning to snow, dramatic cinematic lighting, American flag subtly visible in background. No text, no captions." Need to include [Spezifische Szene aus dem Witz] placeholder? The instruction says: "[Spezifische Szene aus dem Witz]" in the image_prompt. So we need to embed that placeholder? Actually the template includes "[Spezifische Szene aus dem Witz]" as a placeholder to be replaced with specific scene from the joke. So we should include that placeholder in the image_prompt? The instruction: "image_prompt": "Englischer Prompt für Bildgenerierung: A hyper-realistic portrait of a tough, legendary action hero man in his 50s with reddish-brown hair, thick reddish beard, piercing eyes, muscular build. Black sleeveless vest. [Spezifische Szene aus dem Witz]. Photorealistic action movie poster, dramatic cinematic lighting, American flag background. No text, no captions." Thus we need to produce a prompt that includes the placeholder? Or we need to replace it with the specific scene? The instruction says: "image_prompt": "Englischer Prompt für Bildgenerierung: A hyper-realistic portrait ... [Spezifische Szene aus dem Witz]. Photorealistic action movie poster, dramatic cinematic lighting, American flag background. No text, no captions." Thus we need to embed the placeholder exactly as "[Spezifische Szene aus dem Witz]"? Or replace it with the scene description? The example says: "A hyper-realistic portrait ... [Spezifische Szene aus dem Witz]." So we should put the placeholder there, but maybe we need to fill it with the specific scene from the joke. The instruction says: "image_prompt": "Englischer Prompt für Bildgenerierung: A hyper-realistic portrait ... [Spezifische Szene aus dem Witz]. Photorealistic action movie poster, dramatic cinematic lighting, American flag background. No text, no captions." Thus we need to include the placeholder text exactly? Or we need to replace it with the scene? The phrase "Spezifische Szene aus dem Witz" suggests we need to insert the specific scene from the joke. So we should replace that placeholder with a description of the scene. So the image_prompt should contain the scene description in English, describing the specific scene from the joke. So we can write: "Chuck Norris blinking, causing a snowstorm to erupt over a desert landscape, sand turning to snow". That would be the specific scene. Thus image_prompt: "A hyper-realistic portrait of a tough, legendary action hero man in his 50s with reddish-brown hair, thick reddish beard, piercing eyes, muscular build. Black sleeveless vest. Chuck Norris blinking, causing a snowstorm to erupt over a desert landscape, sand turning to snow. Photorealistic action movie poster, dramatic cinematic lighting, American flag background. No text, no captions." That seems fine. Now meta_title max 60 chars: maybe "Chuck Norris lässt Schnee in der Sahara fallen". Count characters: Let's count: "Chuck Norris lässt Schnee in der Sahara fallen" length? Count: C(1)h2u3c4k5 (space6)N7o8r9r10i11s12 (space13)l14ä15s16t17 (space18)S19c20h21n22e23e24 (space25)i26n27 (space28)d29e30r31 (space32)S33a34h35r36a37 (space38)f39a40l41l42e43n44. That's 44 characters, okay under 60. Meta_description max 160 chars: maybe "Ein humorvoller Chuck-Norris-Witz über einen Schneesturm in der Sahara, den er mit einem einzigen Blinzeln erzeugt." Count characters: Let's count roughly: "Ein humorvoller Chuck-Norris-Witz über einen Schneesturm in der Sahara, den er mit einem einzigen Blinzeln erzeugt." That's about 106 characters, okay. Now ensure JSON format, no markdown code block, just raw JSON. Fields: joke_text, headline