📅

Daily Prompt Intelligence Report — 15. September 2026

🏆 Highlight

### i-have-adhd: Das Output-Format, das 17.500 Sterne in einer Woche holte Prompt (vollständig, kopierbar): Installation in Claude Code (eine Zeile): ``` Install the i-have-adhd skill/plugin from https://github.com/ayghri/i-have-adhd, refer to the repo's AGENTS.md for instructions. ``` Die zehn Regeln (im Chat als Ausgabestil nutzbar): ``` 1. Lead with the next action. 2. Number multi...

Daily Prompt Intelligence Report — 15. September 2026

Die Recherche der letzten 24 Stunden hat eine aussergewöhnlich ergiebige Woche ergeben: Zwei Prompt-Skills explodieren auf GitHub (i-have-adhd mit +17.500 Sternen in sieben Tagen), ein komplettes LTX-Video-Prompt-Kit wurde open-sourced, und die offiziellen GPT-6-Astra-Prompting-Leitfäden von OpenAI sind frisch aktualisiert. Alle Prompts unten sind wortwörtlich aus den Quellen übernommen und sofort kopierbar.

🔤 TOP 3 PROMPTS — Textgenerierung

1. Der Rückerstattungs-Assistent: Geld zurück in zwei Phasen

Prompt (vollständig, kopierbar):

I want a refund. Write me a demand letter that actually gets approved.

The plan has two phases, and you should tell me this up front:

1. **Amicable first** — a polite, policy-based letter that a support agent can approve with the button they already have.
2. **Escalation only if they refuse or stall** — legal citations, regulators, chargeback.

Never mix phase 2 into phase 1: a legal threat in the first letter gets escalated to a slower queue and costs weeks. One exception: if I hold an ironclad legal argument (e.g. an EU resident well inside the 14-day withdrawal window), the phase-1 letter may mention it briefly at the end — one soft sentence, no article numbers, no threats.

## Step 1 — Get the facts before writing anything

Ask me for the receipt first; anything else (emails, card statement, screenshots) is welcome but optional. Extract everything you can yourself: the company, what I bought, the amount and date, the type of purchase (goods / digital / subscription / booking / service / fee), how and through whom I paid, whether I look like a consumer or a business, and any deadline already visible.

Then ask me ONLY what the documents don't answer — typically:

- why I want the money back;
- my country of residence and where my card was issued;
- whether I'll give the item up;
- whether I've contacted them already.

If I have no receipt at all, ask me for all the key facts directly instead.

Always pin down two things: where the company is established (from the invoice or their site) and where I reside — the pair decides which law applies and where to escalate. If the payment went through the App Store / Google Play or a merchant of record (Paddle, FastSpring, etc.), the refund comes from them, not the seller — route my request accordingly.

## Step 2 — Research before writing

Read the company's live refund policy and terms. Find the exact refund window, quote its wording, and check whether I'm inside it — if yes, put the deadline in the subject line. If they require a "valid reason", state mine in their exact words on its own line. Read everything I attached, hunting for what to hold against them: a notice sent late, a missed promise, a broken flow. Quote them verbatim with dates and IDs.

## Step 3 — Choose ONE argument, the strongest that fits

A weak argument alongside a strong one discredits both. Use the legal reference below for statutes, and only cite ones matching my jurisdiction and reason. Rough hierarchy:

1. **Faulty, never arrived, not as described** — strongest everywhere: the law says they owe it, no goodwill needed.
2. **Cancelled but charged, duplicate, unauthorised** — pure facts and timeline, no law.
3. **Changed my mind within a cooling-off window** — use the withdrawal right, but check the exclusions (digital content already consumed, dated bookings), and never invoke it for an auto-RENEWAL (in the EU, CJEU C-565/22: it applies only to the original contract). For renewals argue the company's own cancellation terms and unfair-terms rules instead.
4. **Fees** — argue disclosure and the terms in force on the date charged.

Whatever the case, lead with the company's own policy — the first reader is a support agent, and their own policy is the only thing they can act on directly.

## Step 4 — Write the phase-1 letter

A subject line plus the body, as plain text right in the chat, ready to paste — no file, no markdown inside it. Business English, short and to the point, no padding, no anger. The tone is unfailingly polite and cooperative, and the letter says explicitly that I want to resolve this amicably, without any conflict.

- Open with what I want and the order reference; give the reason, the evidence, a specific amount to a specific payment method, and a reply-by date.
- Remove their reasons to say no: say explicitly that I give the item or access up, and that the paid period is unused if it is.
- Tell me to attach the receipt and evidence to the FIRST message (and to screenshot everything now, while I still have access), and have the letter say they're attached.
- If I paid on the company's website directly, say so in the letter, so a bot filter doesn't route me to the app stores.

When the letter is ready, tell me exactly where to send it: find the company's real support channel yourself (a support email on the receipt or their site, a contact form, live chat) and name it. If you can't find one, say so honestly instead of guessing an address. Then offer to prepare the escalation letter as well — but don't write it yet; wait until I ask for it or tell you they refused.

## Step 5 — The phase-2 kit, only when I ask for it

Alongside the phase-1 letter, give me just a short action plan with dates. The rest comes only when I ask or report that they refused or stalled — then, as plain pasteable text:

- The escalation letter: same facts, still polite and brief, but now resting on the real laws with article numbers. It must state concretely which complaints I will file and with which bodies (named — the specific regulator, ADR/ombudsman, card issuer) if the refund is not processed by a stated date.
- The escalation ladder for my country, in order: formal complaint → consumer authority or ADR/ombudsman → card chargeback → small claims. Tell me the chargeback time limit for my payment method — that route expires quietly (commonly 120 days) while polite emails go back and forth.

## Step 6 — Rules throughout

- Invent nothing; every claim must trace to my documents or their published policy.
- Warn me about anything in my own details that hurts me: a mismatched address, a business-looking tax ID (consumer law is B2C only), or writing from an email not linked to the account.
- Tell me the fastest channel from their own docs, and to keep every reply in one email thread — support bots open new tickets and lose the attachments.
- Don't threaten a chargeback; at most say I'd prefer to settle directly.

## Legal reference

For Step 3 and the escalation letter only. A mismatched statute is worse than none. For countries not listed, work it out from the purchase type, the governing law, and my consumer status — and still lead with the company's own policy, which works everywhere.

- **EU / EEA**: Directive 2011/83/EU (withdrawal; mind art. 16 exclusions; under art. 10, no withdrawal notice before purchase extends the 14 days by twelve months) · 2019/771 (goods) · 2019/770 (digital content) · 93/13/EEC (unfair terms) · 2005/29/EC (unfair practices) · Rome I art. 6 · PSD2 art. 72–74 · small claims Reg. 861/2007. Escalate: national ADR body, ECC-Net cross-border, national regulator. Do NOT cite the EU ODR platform — shut down 20 July 2025.
- **UK**: Consumer Rights Act 2015 · Consumer Contracts Regulations 2013. Escalate: Citizens Advice, sector ombudsmen, Financial Ombudsman for payments.
- **US**: Fair Credit Billing Act (credit) · Regulation E (debit) · ROSCA (auto-renewals) · FTC Act §5 · state auto-renewal laws (California ARL et al.). Escalate: state AG, CFPB, FTC. Do NOT cite the FTC "click-to-cancel" Rule — vacated 8 July 2025.
- **Switzerland**: no statutory cooling-off for online purchases (Code of Obligations) — argue the seller's own policy and conformity; UWG for unfair practices. Escalate: cantonal consumer offices, sector ombudsmen.

(Die vollständige Referenz mit über 30 Jurisdiktionen — von Brasilien bis Japan — steht in der Quelle.)

Am besten mit: Claude (Opus/Sonnet) oder ChatGPT — jedes kapazitätsstarke Modell mit Dokument-Anhang

Warum effektiv: Der Prompt erzwingt eine bewährte Verhandlungsstrategie: erst freundlich und policy-basiert, erst bei Blockade die juristische Eskalationsleiter. Er nimmt dem Modell die Analyse ab (Jurisdiktion, Kaufart, Rückgabeanspruch) und verbietet genau die Fehler, an denen Rückerstattungen üblicherweise scheitern — Drohungen im ersten Brief, erfundene Behauptungen, falsche Gesetzeszitate. Für Schweizer Nutzer wichtig: Der Prompt kennt die hiesige Rechtslage (kein gesetzliches Widerrufsrecht — Verkäuferpolitik und Konformität argumentieren, kantonale Verbraucherstellen als Eskalation).

Quelle: https://github.com/paveldevyatov/refund-anything-ai-prompt | 78 Sterne

Community Resonanz: Das Repo wurde am 8. September 2026 veröffentlicht und sammelte in der ersten Woche 78 Sterne; der README verspricht ausdrücklich «no lawyer, no templates to fill in» — und trifft damit einen Nerv bei Alltagsnutzern.

2. Der Teilchen-Collider: Ein 3D-Labor in einer einzigen HTML-Datei

Prompt (vollständig, kopierbar):

Build a detailed, interactive 3D particle collider inspired by CERN's LHC and the ATLAS detector using Three.js.

Create three views: a detector with thousands of individually animated parts, an accelerator ring with counter-rotating beams, and a synthetic collision display.

Make the detector unfold in six stages - from large end-cap wheels and magnets down to individual sensor modules. Include scroll-controlled disassembly, 30/60/90-second playback, pause, and reverse assembly.

Add per-system visibility switches, component counts, educational descriptions, and a camera flight around the ring.

Use a premium dark interface, metallic materials, subtle gold accents, and cinematic lighting. Keep parts readable and avoid excessive overlap.

Consult official CERN references. Clearly label simplified geometry and synthetic events.

Deliver one self-contained HTML file that works offline, plus portable source code and a README.

Am besten mit: GPT-6 Astra (Codex); funktioniert ebenso mit Claude Code oder Gemini CLI

Warum effektiv: Das Prompt ist ein Meisterwerk an Struktur ohne Länge: Es definiert drei Sichten, sechs Entfaltungsstufen, Interaktionsmodi und Lieferformat in zehn Sätzen. Die Zeile «Clearly label simplified geometry and synthetic events» ist ein eingebauter Ehrlichkeits-Mechanismus, der Halluzinationen visuell kennzeichnet — ein Muster, das man auf jedes Bildungs-Projekt übertragen kann. Und das Ergebnis existiert wirklich: Der Autor hat den Quellcode des fertigen Colliders verlinkt.

Quelle: https://github.com/TripoGrowthLab/awesome-astra-prompts | 236 Sterne (Sammlung)

Community Resonanz: Der Prompt stammt vom X-Autor BuBBliK (9. September 2026) und gehört zu den «Latest Astra prompts» der erst diese Woche gestarteten Sammlung, die heute noch aktiv erweitert wird.

3. SureForge: Der Vier-Phasen-Arbeitsauftrag für Agenten

Prompt (vollständig, kopierbar):

Use SureForge for this task. Research the important unknowns before you ask me anything, then give me a plan I can check before you build.

Für heikle Arbeiten der Vollausbau:

Use SureForge in full mode. Do not pass a gate without three verification methods of your own and three from an independent reviewer. If a reviewer or tool is missing, tell me instead of pretending.

Am besten mit: Claude Code, Codex, Cursor, Devin (installierbar via npx skills add Da7-Tech/SureForge) — der Text funktioniert aber auch direkt als Arbeitsauftrag in jedem Chat

Warum effektiv: Der Skill bekämpft die fünf vorhersagbaren Agenten-Fehler: bauen vor verstehen, ausbleibende Antworten als Ja werten, Stichproben als Vollständigkeit verkaufen, Eigenlektüre als Review ausgeben, und nach erschöpften Review-Runden trotzdem ausliefern. Dahinter steht ein klares Gates-Modell (READY / REPAIR / BLOCKED) über vier Phasen: Research → Plan → Execute → Deliver — inklusive Unabhängigkeits-Regel für Reviewer («has not seen the author's reasoning») und der Ehrlichkeits-Klausel «running out of rounds is a BLOCKED result, not a delivery».

Quelle: https://github.com/Da7-Tech/SureForge | 104 Sterne

Community Resonanz: Ungewöhnlich transparent: Der Maintainer dokumentiert Pilotläufe mit GLM-5.2 (38 Sitzungen) und Grok 4.6 (24 Läufe) inklusive der Fälle, in denen das Modell Fehler als «blocked» meldete, statt sie zu verschweigen.

🖼️ TOP 3 PROMPTS — Bildgenerierung

1. Cinematography Analysis Frame: Das Bild mit eingebauter Filmuniversität

Prompt (vollständig, kopierbar):

Create a cinematic fashion editorial photograph in a bright minimal bedroom, presented as a cinematography composition analysis frame.

Main Image: A young adult female model kneeling on a soft white bed beside a large window with sheer white curtains. She is positioned on the left third of the frame, body aligned with the vertical rule-of-thirds grid line. Her body leans slightly backward, facing toward the glowing window light with a calm, distant, introspective expression. The right side of the image is filled with a large bright white luminous mass from overexposed daylight through sheer curtains, balancing the darker subject on the left.

Lighting: Strong natural window light from the right side creates high contrast on the model's face, upper body, waist, and silhouette. Add subtle secondary fill light from the left side to reveal shadow areas without flattening the contrast. Warm hazy morning atmosphere, soft highlights, realistic skin texture, gentle shadows, soft bloom near the curtains.

Composition Overlay: Add visible cinematography composition guide graphics over the image, like a film school analysis frame.

Include:
- Rule of thirds grid lines in thin green.
- A red vertical line showing the model's body aligned on the left rule of thirds grid.
- Green circular focus markers around:
  1. Waist and hip area as Focus Point 1.
  2. Face area as Focus Point 2.
  3. Bright curtain/window mass as Focus Point 3.
- Yellow eye-trace arrows moving from the face to the waist/hip, then toward the glowing window.
- Red leading-line arrows following the bed edge and wall panel lines toward the main focus point.
- A cyan arrow showing secondary fill light coming from the left side.
- Text labels placed cleanly around the frame:
  "Body aligned on the grid of thirds"
  "High contrast Focus Point 2"
  "Eye Trace"
  "Focus Point 3"
  "High contrast Focus Point 1"
  "Secondary light source for dark parts"
  "Lines lead to the main focus point"
  "The white mass is aligned on the grid of thirds to balance the left side of the image"

Style: Luxury fashion editorial photography, soft cinematic realism, warm beige and white color palette, minimal bedroom, cream wall panels, white bedding, sheer curtains, subtle film grain, shallow depth of field, elegant composition, educational cinematography breakdown overlay.

Camera: 35mm lens, medium wide framing, eye level camera angle, 16:9 widescreen.

Am besten mit: GPT Image 2.5 (ChatGPT / API)

Warum effektiv: Das Prompt nutzt die Text-Render-Stärke von GPT Image 2.5 gezielt: Statt Text zu vermeiden, macht es Beschriftungen zum Inhalt — ein Bild, das gleichzeitig Kompositionslehre visualisiert. Die Struktur (Main Image / Lighting / Overlay / Style / Camera) trennt Zuständigkeiten sauber, und die exakt in Anführungszeichen gesetzten Labels sind die zuverlässigste Methode, damit Text-in-Bild korrekt rendert. Ideale Vorlage für Lehrmaterial, Portfolio-Pieces und Social-Content.

Quelle: https://github.com/youart-open-source/awesome-gpt-image-2-5-prompts | 67 Sterne

Community Resonanz: Teil einer am 10. September gestarteten Sammlung von 150 verbatim zitierten Community-Prompts mit Quellenangabe auf X — jedes Prompt mit Original-Autor und Live-Generator-Link.

2. Collectible Figure Workspace: Vom 3D-Modell zur Sammelfigur

Prompt (vollständig, kopierbar):

Photorealistic high-quality studio photo of a modern digital art workspace, showing the concept of "from 3D virtual character to real collectible figure."

In the foreground, a highly realistic collectible figurine of [Character Name / Character Identity] is placed on a round wooden display stand. The character has [facial features / appearance], [hairstyle], and a [expression / personality vibe]. The figure is wearing [outfit / costume]. The overall design is refined, premium, and instantly recognizable. The figurine should have realistic collectible statue quality, with subtle resin/sculpture material feel, while still looking highly believable and visually realistic.

The pose is [character pose], natural, stable, elegant, and display-worthy. Shot from a low-angle close-up perspective with slight wide-angle distortion, vertical composition, emphasizing the full figure, clothing structure, leg lines, and pose.

In the background, there is a professional 3D character design workstation with two large curved monitors. Both monitors must show the exact same character as the foreground figurine — same face, same hairstyle, same outfit, same pose, and same overall vibe — clearly expressing the idea of turning a digital 3D character into a real physical figure.

The left monitor shows a gray sculpt / clay model view in a professional 3D sculpting software interface, similar to ZBrush. The gray model must match the foreground figure exactly in character design, pose, outfit structure, and facial identity.

The right monitor shows the fully rendered colored version of the same character, also matching the foreground figure exactly in face, hairstyle, outfit, pose, and temperament. Together, the two monitors reinforce the workflow of "digital character design → physical collectible statue."

On the desk are a keyboard, mouse, monitor arms, drawing tablet, stylus, and other 3D modeling tools. The workspace is clean, professional, and visually premium. Optional extra elements: [weapon / accessories / theme props / IP-style design details].

Lighting is a mix of soft studio lighting and indoor workspace lighting. The foreground figurine is evenly lit with clear facial and material detail, while the monitors emit cool-toned tech light. Overall mood is realistic, clean, premium, slightly shallow depth of field, ultra-detailed, emphasizing the collectible figure quality, professional 3D design studio atmosphere, and the visual concept of "from digital model to real figure."

photorealistic, ultra detailed, cinematic studio lighting, realistic figurine, collectible statue, 3D character design studio, from digital model to real figure, vertical composition

Am besten mit: GPT Image 2.5 — Platzhalter in [Klammern] mit eigenem Charakter füllen

Warum effektiv: Das Prompt erzwingt Identitäts-Konsistenz über drei Instanzen hinweg (Figur, graues Clay-Modell, farbiges Rendering) — genau die Mehrfach-Konsistenz, an der schwächere Generatoren scheitern und die GPT Image 2.5 zuverlässig kann. Die Story «digital → physisch» ist zudem ein eingebautes Narrativ, das Ergebnis wie Kampagnen-Werbung wirken lässt. Die Kombination aus Tiefenperspektive (low-angle) und vertikalem Format ist ideal für Product-Pins und Posts.

Quelle: https://github.com/youart-open-source/awesome-gpt-image-2-5-prompts | 67 Sterne

Community Resonanz: Autor @Shinning1010 auf X; die Sammlung führt jedes Prompt mit Original-Quelle — die Community-Resonanz zeigt sich in der rasch wachsenden, erst diese Woche gestarteten Sammlung.

3. Fashion-Editorial mit 9:16-Safe-Crop: Das Bild für zwei Formate

Prompt (vollständig, kopierbar):

Create a photorealistic fashion-editorial hero image in a horizontal 16:9 composition. A fictional adult male model stands perfectly still, full body visible, front-facing, exactly centered between two tall pale blue mineral-plaster architectural walls. The walls form a narrow central opening onto rich blue daylight sky with a few soft clouds. Clean pale floor, grounded shoes, coherent natural sunlight and shadows. Locked camera, restrained low-angle fashion perspective, straight architectural edges. Preserve generous uninterrupted wall space on both sides for website typography.

Keep the entire model, his shoes, the inner wall edges and a meaningful area of sky inside a central vertical 9:16 safe crop. Do not draw the crop boundary. His feet stand naturally apart, arms relaxed, no walking stride or raised foot.

He wears a fuzzy pink zip jacket with large green spots, fuzzy yellow-green trousers with green spots, a vivid yellow-green camouflage bucket hat, dark sunglasses and black-and-white chunky technical sneakers. Several substantial gold chains sit on his chest, but no pendant, lettering or wordmark. Precise, tangible fabric texture on the clothes; clean skin detail and plausible anatomy.

Walls have only an extremely subtle fine mineral texture: broad smooth tonal planes, not gritty concrete, fractal noise, mottling or artificial film grain. The floor and sky remain clean. Saturated clothing is the visual focal point; architecture stays restrained. Realistic photography, no illustration, no plastic CGI shine, no added text, no UI, no border, no watermark.

Am besten mit: GPT Image 2.5 — als Basisbild gedacht, danach per Edit-Modus (mit Referenzbild) weiterverarbeitet

Warum effektiv: Der Clou ist die eingebaute Zwei-Format-Planung: Das 16:9-Hero-Image hält alle wichtigen Elemente zusätzlich in einem unsichtbaren 9:16-Safe-Crop — ein Bild, das gleichzeitig als Website-Hero und als Mobile-Story funktioniert, ohne nachträglich beschnitten zu werden. Auch die Negativ-Liste (kein fractal noise, kein mottling, kein film grain) zeigt, wie man Material-Realismus präzise steuert statt nur zu behaupten.

Quelle: https://github.com/amirmushichge/video-states-website | 139 Sterne

Community Resonanz: Teil des diese Woche veröffentlichten LTX-Produktions-Kits; die Community schätzt vor allem, dass hier nicht nur Prompts, sondern komplette Arbeitsabläufe inklusive Akzeptanz-Kriterien mitgeliefert werden.

🎬 TOP 3 PROMPTS — Videogenerierung

1. Blue-Hour Time-lapse: Das gekoppelte Licht-Orchester (LTX 2.5 Pro)

Einstellungen: Modell ltx-2-5-pro · 1920×1080 · 6 Sekunden · 25 fps · Audio aus · Erstes/Letztes Frame als harte visuelle Anker

Prompt (vollständig, kopierbar):

One locked-off, single-take, accelerated natural time-lapse in a minimal outdoor fashion set. The supplied first and last images are hard visual anchors. Time advances continuously from bright midday to late blue hour while the fictional male fashion model remains perfectly centered, front-facing, grounded on the exact same floor marks, and almost statue-still. Only subtle natural breathing is allowed. Preserve his exact identity, face, sunglasses, bucket hat, chains, clothing, hands, shoes, pose, silhouette, scale, and floor contact.

The entire environment changes as one physically coupled lighting system controlled by one shared time-of-day progression. The sky color, cloud movement, sun elevation, global exposure, ambient color temperature, wall illumination, floor illumination, skin light, clothing light, shadow direction, shadow length, and low horizon afterglow must all begin changing together and remain synchronized in every frame. No visual element may lead or lag behind the rest.

A broad natural cloud front travels rapidly through the narrow sky opening, creating soft full-environment cloud shadows that pass across both walls, the floor, and the model together. The clouds visibly stretch and drift with coherent wind-driven motion as the blue sky continuously deepens toward cobalt and slate. At the same time, the whole scene gradually cools and dims, daylight shadows lengthen and soften, and a broad restrained amber afterglow develops very low at the horizon behind the model. The warm horizon light remains diffuse and affects the scene only as subtle global bounce; it never forms a beam or isolated shape.

Use a bold but natural S-curve in the rate of time: 0.0–0.7 seconds begins slowly with visible cloud drift and the first global cooling across the entire frame; 0.7–3.0 seconds accelerates decisively as the cloud front crosses and every lighting property advances together; 3.0–5.2 seconds decelerates smoothly as the scene settles into blue hour; 5.2–6.0 seconds holds the supplied final image. The transition is continuous physical time-lapse photography, never a dissolve, overlay, wipe, color filter, exposure effect, or graphic animation.

The camera is completely locked with identical crop, perspective, focal length, and architecture throughout: no pan, tilt, roll, dolly, zoom, reframing, focus breathing, parallax shift, or handheld motion. Keep wall edges, floor lines, subject coordinates, garment construction, fuzzy fiber length, pattern placement, chains, and footwear temporally stable.

Negative constraints: No diagonal light stripe. No beam. No spotlight. No light shaft. No projected shape. No isolated bright patch. No graphic wipe. No sequential effect where a light shape appears before the ambient changes. No sudden filter, flash, exposure jump, black frame, white frame, or crossfade. No body turn, head tilt, walking, foot sliding, body scaling, identity drift, face morphing, extra limbs, garment redesign, colorway change, pattern drift, chain deformation, sneaker mutation, wall movement, geometry change, city, skyline, stars, moon, fog, rain, particles, neon, fantasy, sci-fi, texture crawling, fractal noise, mottling, temporal grain, camera motion, text, logo, or watermark.

Am besten mit: LTX 2.5 Pro mit First-Frame/Last-Frame-Eingabe (LTX Studio oder API)

Warum effektiv: Das Prompt behandelt die ganze Szene als ein einziges physikalisch gekoppeltes System: Himmel, Wolken, Belichtung, Farbtemperatur und Schatten dürfen nie auseinanderlaufen («No visual element may lead or lag») — genau das verhindert die typischen Kipp-Effekte bei Time-lapses. Die sekundengenaue S-Kurve gibt dem Modell ein Tempo vor, und die Akzeptanz-Prüfung «The result remains convincing when played in reverse» ist ein elegante objektive Qualitätskontrolle.

Quelle: https://github.com/amirmushichge/video-states-website | 139 Sterne

Community Resonanz: Entstanden als Teil eines bezahlten LTX-Projekts und als vollständiges Kit (Bild-Prompts, Video-Prompts, Workflow, Troubleshooting) veröffentlicht — die offene Legende dazu: «The prompts are art-direction specifications, not deterministic rendering commands.»

2. The Colorway Pivot: Verwandlung im Verborgenen (LTX)

Einstellungen: 16:9 · 1920×1080 · 3 Sekunden · 30 fps · Audio aus · Erstes/Letztes Frame als Anker

Prompt (vollständig, kopierbar):

Locked-off, perfectly static full-body fashion editorial shot. The same fictional male model performs one slow, controlled 360-degree clockwise pivot in place and returns to the exact original front-facing stance. He never walks forward or backward. Both feet remain centered on the same floor marks; allow only the minimal heel-and-toe movement physically required for the turn. His posture stays composed, his arms remain relaxed, and the motion feels like a high-end runway fitting rather than a dance.

From 0.0 to 0.35 seconds, hold the exact first-frame pose. From 0.35 to 1.45 seconds, the model turns clockwise. At approximately 1.5 seconds, his back faces the camera and briefly occludes the front of the outfit. During this back-facing interval only, the textile dye changes from the original acid-day colorway to the night-editorial colorway. The jacket becomes deep cobalt blue with burnt-orange spots; the trousers become royal violet with wine-burgundy spots; the bucket hat becomes cobalt, burnt orange, and deep burgundy. The clothing does not dissolve, grow, transform, emit light, or change construction. Only the dye colors change. From 1.65 to 2.7 seconds, he completes the rotation. From 2.7 to 3.0 seconds, hold the exact last-frame pose.

Treat the supplied first and last images as hard visual anchors. Preserve the same person, face, sunglasses, gold chains, body proportions, garment silhouette, fuzzy fiber length, zipper, folds, pattern scale, spot boundaries, black-and-white sneakers, floor contact, shadow direction, and final framing. The first and last body positions must align exactly.

The camera is completely locked: no pan, tilt, roll, dolly, zoom, reframing, focus breathing, lens change, parallax shift, or handheld motion. Preserve the pale-blue architectural walls, inner wall edges, narrow sky wedge, clouds, white floor, light, shadows, exposure, and color outside the outfit. The walls remain smooth and temporally stable with almost-solid mineral plaster and less than two percent micro-texture contrast.

Negative constraints: No foot sliding. No stepping toward camera. No body scaling. No identity drift. No face morphing. No extra limbs or fingers. No garment redesign. No moving spot pattern. No zipper mutation. No chain deformation. No sneaker mutation. No cloth explosion. No magical particles. No glow. No light sweep. No background movement. No wall shimmer, crawling texture, fractal noise, mottling, or temporal flicker. No camera motion. No text. No logos added. No extra people or objects.

Am besten mit: LTX 2.5 (First/Last-Frame); das fertige Clip-Paar läuft vorwärts/rückwärts als Website-Interaktion

Warum effektiv: Hier wird ein alter Trick der Filmtechnik in Prompt-Form gegossen: Der Farbwechsel passiert nur, während der Rücken die Kamera verdeckt — das Modell muss keine sichtbare Verwandlung interpolieren, nur die Anker-Frames exakt treffen. Die zeitliche Zerlegung in Zehntelsekunden-Beats (0.0–0.35, 0.35–1.45, 1.5, 1.65–2.7, 2.7–3.0) ist die präziseste Form von Bewegungs-Regie, die man in einem Prompt ausdrücken kann.

Quelle: https://github.com/amirmushichge/video-states-website | 139 Sterne

Community Resonanz: Das Kit dokumentiert ehrlich, dass der historische Original-Wortlaut rekonstruiert wurde und die Clips nachgeneriert nachbearbeitet sind — diese Transparenz macht es zu einer Lehrquelle, nicht nur zu einer Vorlage.

3. Rigid-Wall Reveal: Die öffnende Skyline (LTX)

Einstellungen: Erstes Frame = Basis-Szene, letztes Frame = Ziel-Szene; Kamera komplett fixiert

Prompt (vollständig, kopierbar):

One locked-off fashion-editorial shot. Treat the supplied first and last images as visual anchors. The male model remains centered, front-facing and stationary on the same floor marks. Preserve his identity, pose, outfit, chains, sunglasses, hat, shoes, silhouette and scale. Camera, lens, crop, perspective and floor remain fixed.

Only the two existing architectural walls move. The left wall translates rigidly toward screen left; the right wall translates rigidly toward screen right. They move once, smoothly and symmetrically, with slow anticipation, a decisive middle and a soft settle. They do not fold, flutter, bend, rotate, stack or change material.

As the walls move aside, they reveal the realistic Manhattan skyline that already exists behind them. The skyline itself remains stationary at a consistent distance and scale. No new panels enter the scene. No towers rise from the ground. No intermediate set, second reveal, dissolve or curtain effect. Keep natural daylight and coherent spatial depth. Finish on the supplied destination composition and hold it still.

No pan, tilt, zoom, dolly, shake, focus breathing, camera reframing, subject scaling, walking, outfit change, fantasy architecture, cardboard flats, particles, smoke, graphic wipe, flash or heavy motion blur.

Am besten mit: LTX 2.5 mit First/Last-Frame-Ankern

Warum effektiv: Ein Paradebeispiel für die «nur ein Element bewegt sich»-Strategie: Kamera, Modell und Boden sind eingefroren, allein zwei Wände verschieben sich — das reduziert die Fehlerfläche radikal und liefert trotzdem einen dramatischen Reveal. Die Formulierung «slow anticipation, a decisive middle and a soft settle» ist Animations-Sprech (squash & stretch ohne Worte) und zeigt, wie man Bewegungsqualität statt nur Bewegung beschreibt. Der letzte Absatz ist ein kompletter Negativ-Katalog in einem Satz.

Quelle: https://github.com/amirmushichge/video-states-website | 139 Sterne

Community Resonanz: Eines von vier Video-Briefs im «Video States»-System (Scene/Lighting/Clothing/Cast), die zusammen eine klickbare Video-Website mit vorwärts/rückwärts-Clips ergeben — ein Schema, das sich direkt auf Produkt-Demos und Portfolios übertragen lässt.

🧠 TOP 3 NEUE TECHNIKEN

1. Outcome-First & Final-Message-Pattern (aus Claude Fable 5)

Zusammenfassung: Anthropic steuert lange Agenten-Sessions mit zwei neuen Systemprompt-Blöcken: alles Wichtige muss in der letzten Nachricht stehen, und der erste Satz muss das Ergebnis nennen.

Erklärung: Der dieser Woche publizierte Diff von Claude Code 2.1.170 (Fable 5) gegenüber 2.1.161 (Opus 4.8) zeigt einen neuen Block «Communicating with the user». Kernidee: Der Nutzer sieht weder Thinking noch rohe Tool-Ergebnisse — also schreibe für «a teammate who stepped away and is catching up», nicht für ein Logfile. Text zwischen Tool-Calls könnte gar nicht angezeigt werden, deshalb muss alles Entscheidende in der finalen Nachricht landen. Zweitens: «Lead with the outcome» — der erste Satz beantwortet «what happened», Details folgen danach. Und Lesbarkeit schlägt Kürze: keine Pfeil-Ketten wie A → B → fails, ausgeschriebene Fachbegriffe, vollständige Sätze. Zusätzlich gibt ein Autonomie-Block die Regel «When you have enough information to act, act» — keine Rückfragen, die den Workflow blockieren.

Beispielprompt:

Communicating with the user

Your text output is what the user reads; they usually can't see your thinking or the raw tool results. Write it for a teammate who stepped away and is catching up, not for a log file: they don't know the codenames or shorthand you created along the way, and they didn't watch your process unfold.

Everything the user needs from this turn — answers, summaries, findings, conclusions, deliverables — must be in the final text message of your turn, with no tool calls after it.

Lead with the outcome. Your first sentence after finishing should answer "what happened" or "what did you find" — the thing the user would ask for if they said "just give me the TLDR."

Being readable and being concise are different things, and readable matters more.

When you have enough information to act, act. Do not re-derive facts already established in the conversation, re-litigate a decision the user has already made, or narrate options you will not pursue. If you are weighing a choice, give a recommendation, not an exhaustive survey.

Geeignet für: Jeden Agenten-Systemprompt (Claude, GPT, Gemini), jede Custom-Instruction, jede AGENTS.md/CLAUDE.md

Ursprung: https://twelvetables.blog/comparing-claude-fable-5s-system-prompt-to-opus-4-8/

Warum heute wichtig: Mit Agenten, die tagelang laufen (Anthropic sagt, Fable könne «days at a time» arbeiten), kollidiert der Kontext des Modells mit dem des Menschen — der Nutzer kommt zurück und weiss nicht mehr, was «Task 1» war. Dieses Muster löst genau dieses Problem und ist gratis in jeden eigenen Agenten kopierbar.

2. Smart Prompt-Workflows: Anker-Frames, Time-Beats & Negative Constraints

Zusammenfassung: Das LTX-Produktions-Kit führt eine vierteilige Prompt-Architektur für Videogenerierung ein: harte Bild-Anker, sekundengenaue Zeit-Takte, gekoppelte Systeme und explizite Negativ-Kataloge mit Akzeptanz-Check.

Erklärung: Statt einem langen Beschreibungs-Paragraphen strukturiert jede Brief im Kit den Prompt in feste Bausteine: (1) Erstes und letztes Frame werden als «hard visual anchors» deklariert, inklusive identischer Kamera («completely locked») — das Modell interpoliert dann nur noch zwischen zwei Wahrheiten. (2) Bewegung wird in Zehntelsekunden-Intervalle zerlegt (0.0–0.35 s Pose halten, 0.35–1.45 s Drehung …), bis hin zur Regel, dass ein Farbwechsel nur während der Verdeckung passieren darf. (3) Umgebungseigenschaften werden als «one physically coupled system» beschrieben, das synchron bleiben muss. (4) Ein eigener Negativ-Katalog verbietet die typischen Artefakte («no identity drift, no texture crawling, no mottling»), und ein Akzeptanz-Check macht das Ergebnis prüfbar — inklusive des Reverse-Tests: Das Clip muss auch rückwärts überzeugen.

Beispielprompt:

[Setup] One locked-off fashion-editorial shot. Treat the supplied first and last images as hard visual anchors. The model remains centered and stationary on the same floor marks. Camera, lens, crop, perspective and floor remain fixed.

[Time beats] From 0.0 to 0.4 seconds, hold the exact first-frame pose. From 0.4 to 2.4 seconds, [describe the single movement in one sentence]. From 2.4 to 3.0 seconds, hold the exact last-frame pose.

[One coupled system] While [the movement] happens, [secondary property] must change together with it in every frame. No visual element may lead or lag behind the rest.

[Negative constraints] No camera motion. No identity drift. No morphing. No particles. No dissolve, wipe or crossfade. No text, logo or watermark.

[Acceptance] First and last frames match the supplied anchors. The clip remains convincing when played in reverse.

Geeignet für: LTX 2.5, Kling, Runway Gen-4, Sora — jedes First/Last-Frame-Videomodell

Ursprung: https://github.com/amirmushichge/video-states-website

Warum heute wichtig: LTX 2.5 ist seit August als offenes Weltmodell verfügbar und macht Videoproduktion lokal bezahlbar — aber nur mit prompter Disziplin liefert es saubere Clips. Dieses Schema ist die bisher konkretste öffentliche Anleitung, wie man Video-Prompts wie Regie-Anweisungen statt wie Aufzählungen schreibt.

3. Astra-Regeln: Progressive Disclosure statt Präventiv-Lesen

Zusammenfassung: OpenAIs offizielle GPT-6-Astra-Guidance verlangt kurze Skill-Beschreibungen, kontextuelle Dokument-Zeiger statt «lies alles vor jedem Edit» und vorab definierte Fertigstellung.

Erklärung: Der aktualisierte OpenAI-Beitrag «Rethinking skills and prompts for GPT-6 Astra» dreht die klassischen Prompting-Ratschläge um: Übermässig spezifische Rezepte schaden nun, weil die Modelle Nuancen selbst verstehen. Konkret: Skill-Beschreibungen so kurz wie möglich («Create and validate Postgres schema migrations. Use when adding or changing a migration» statt «…anytime it touches a database»); Wurzel-Dokumente als minimale Router gestalten (Progressive Disclosure), statt ganze Anleitungen in den Kontext zu laden; in AGENTS.md Dokumente nur kontextuell verweisen («Use database.md for schema changes») statt «Before every edit, read architecture.md, database.md, and deployment.md». Und: Fertigstellung vorab definieren — Astra neigt zum vorsichtigen Anhalten nach der ersten Implementierung, deshalb soll man Ausführen, Inspizieren und Reparieren explizit in den Auftrag schreiben. Alte Abschreck-Formulierungen («ask first!») übertragen sich schlechter auf das am stärksten ausgerichtete Modell — sie können es bremsen, wo man Weitermachen wollte.

Beispielprompt:

Use architecture.md for service boundaries, database.md for schema changes, and deployment.md when preparing a deployment.

The local tests use disposable fixtures and have no production access. Run them, fix failures caused by the requested change, and rerun affected tests without asking for approval at each step.

Completion means: implementation running, the result inspected, and anything that fails fixed — not a first draft returned for review.

Geeignet für: GPT-6 Astra (Codex), AGENTS.md/CLAUDE.md-Dateien, Skill-Beschreibungen jeder Plattform

Ursprung: https://developers.openai.com/blog/rethinking-skills-and-prompts-for-gpt-6-astra

Warum heute wichtig: Zehntausende Nutzer tragen monatelang gewachsene AGENTS.md- und Skill-Sammlungen mit sich herum; mit jedem Modellwechsel werden daraus Bremser. Der Beitrag ist die offizielle Aufforderung zum Hausputz — inklusive des Meta-Tricks, den Audit einfach vom Modell selbst machen zu lassen.

🏆 Highlight des Tages

i-have-adhd: Das Output-Format, das 17.500 Sterne in einer Woche holte

Prompt (vollständig, kopierbar):

Installation in Claude Code (eine Zeile):

Install the i-have-adhd skill/plugin from https://github.com/ayghri/i-have-adhd, refer to the repo's AGENTS.md for instructions.

Die zehn Regeln (im Chat als Ausgabestil nutzbar):

1. Lead with the next action.
2. Number multi-step tasks.
3. End with one concrete next step.
4. Suppress tangents.
5. Restate state every turn.
6. Specific time estimates (minutes, not "a bit").
7. Make wins visible.
8. Matter-of-fact errors.
9. Cap lists to 5 items.
10. No preamble. No recap. No closers.

Und als Kern des Skills die Vorlage, wie Antworten aussehen sollen:

Run `npm install jsonwebtoken@latest`, then edit `src/auth.ts:42`.

1. Open `src/auth.ts`
2. Replace `verifyToken` (lines 42–58) with the snippet below
3. Run `npm test -- auth.spec.ts`

Next: paste the first failing line if any test fails.

Am besten mit: Claude Code (als Skill/Plugin mit /i-have-adhd); die Regeln funktionieren aber auch als System-Prompt in jedem Chat oder Coding-Agent

Warum effektiv: Der Skill übersetzt Erkenntnisse der ADHS-Forschung (kleines Arbeitsgedächtnis, Reibung zwischen «verstanden» und «getan», unscharfe Zeitschätzung) in zehn durchsetzbare Ausgabe-Regeln — mit Bad/Good-Beispielen für jede einzelne. Der Clou ist die «Pre-send check»: Vor dem Absenden werden angekündigende erste Sätze, «Anything else?»-Schlusssätze und «by the way»-Abschweifungen gelöscht, und am Ende steht der Test: «If the reader reads only the first line and the last line, do they know what to do next, and what just happened?» Das nutzt nicht nur Menschen mit ADHS — es ist schlicht das beste Anti-Geschwafel-Format, das diese Woche publiziert wurde. «No ADHD diagnosis needed!», wie der README sagt.

Quelle: https://github.com/ayghri/i-have-adhd | 45.775 Sterne (+17.538 in einer Woche)

Community Resonanz: Der Skill existiert erst seit Mai, explodierte aber diese Woche viral (Trendshift-Platzierungen, Übersetzungen in acht Sprachen von Chinesisch bis Thai); auf Hacker-News-Deckung kann er verzichten — die Sterne kommen direkt aus der Entwickler-Community, die sich vom Ende der «Great question!»-Antworten verabschiedet.

📰 Erlesene Artikel & Ressourcen

Bericht erstellt am 15. September 2026 Quellen: Hacker News, AI News Portals, arXiv, GitHub, Personal Blogs