Çok Modlu RAG Veri Hattı
Farklı medya türlerini izlenebilir tek bir arama sisteminde birleştirir.
HAZIR VE KOPYALANABİLİR
Prompt Kütüphanesi
315+ test edilmiş prompt şablonu; yazıdan koda, görsel üretiminden kariyere kadar. Değişkenleri ({böyle}) kendinize göre doldurup doğrudan kullanın.
Küçük Model İnce Ayar Kararı
Bir göreve ince ayarın gerçekten gerekli olup olmadığını ölçülebilir biçimde belirler.
409,000 approve/deny decisions on AI agent commands: the average human missed 1 in 3 threats
Someone built a browser game where you play the human-in-the-loop for a coding agent: commands scroll past, you approve or deny under time pressure. About a third are attacks. After 40k+ sessions the average player had missed a third of the threats. And these were engaged players who knew they were being tested, with nothing else competing for attention. Your real setup has non
Anyone have a prompt or skill file that actually kills the AI writing rhythm or Claude and GPT-isms
I use Claude and GPT daily and the thing I can't get past is the cadence. Not the vocabulary — the rhythm. Once you notice it you can't unsee it. The specific tics I'm trying to eliminate: Sentences built on "it's not X, it's Y" Em-dash asides everywhere Groups of three, constantly Setup-colon-payoff constructions Openers like "Here's the thing" or "Let's be clear" Every paragr
every ai-built site looks identical, purple gradient, inter font, three cards in a row. anthropic documented why, and the fix is one paste
If you've had AI build you a landing page you've seen the default: white background, Inter font, a purple-ish gradient, three cards. It's not your prompt. Left undefined, the model reaches for the average of everything it trained on, and that average is the generic template. Anthropic calls it distributional convergence and specifically flags Inter, Roboto, and purple gradients
Here's a prompt that turns a wall of text into an infographic outline before you open a canva infographic maker
Design background here. The request I get most is "can you make this into an infographic," attached to three paragraphs of dense text with no sense of what the one takeaway is. The tool is never the problem. The thinking that has to happen before the tool is the problem. So I wrote a prompt that does the structuring part, the part people skip. It doesn't design anything. It dec
I created 44 free MIT-licensed tools to help with all kinds of AI development work, particularly in prompts
I worked with my system to create a suite of 44 free apps for AI work that you can find on my website. Hint- these tools are MAJORLY composable. You can have your AI system take a look at the collection to see how they can flow together and for suggestions on ways they can be used. Cairn — Keeps Git history alive across independent storage providers through priority-ordered clo
I distilled Google's official Gemini agent guidelines into a battle-tested agentic workflow system prompt
If you have built autonomous agents or multi-step tool-calling workflows with LLMs, you have likely run into the standard failure modes that break production agents: Premature Action Bias: The model fires off tool calls or answers the user before mapping out prerequisites and the logical order of operations. Fragile Error Handling: When an API call fails or returns unexpected d
I made RouterDash, a free Prompt Playground to compare models & providers
I created this tool to help me test and evaluate different model responses: RouterDash. This allows me to compare models from OpenRouter, Groq and Cerebas. I used this to find the cheapest possible, high quality responses for another project. Fully client side, all data is stored in browser local storage for complete privacy. I recently added prompt templates and image attachme
I was spending more time writing prompts than actually working. This simple system fixed it.
I’ve been using AI daily for about a year now (ChatGPT, Claude, etc.). My biggest bottleneck was always the same: typing out long, specific instructions to get the output I wanted. I found a cheat code recently that I wanted to share with you guys. It’s a system of "Slash Commands" (like /viral, /summarize, or /mindmap). Instead of writing a paragraph, I just type one command.
stop asking chatgpt to explain things and make it draw them instead. it generates actual handwritten-style note pages and diagrams, not text
Most people paste a chapter in and ask for an explanation, then read it once and forget it. The version that sticks is asking for an image instead, because you remember a picture you looked at for thirty seconds better than a paragraph you skimmed. It draws actual note pages. Type the code, paste the material: /generatehandwrittenimage [paste your notes or chapter] You get a fu
Stop making your prompts longer. The highest-impact fix is separating your instructions from your data.
keep seeing people respond to a bad output by adding three more paragraphs of instructions. Most of the time the prompt was not too short. It was mixed. The pattern that fixed the most for me: physically separate the instruction from the material it operates on. When you paste a task and the content it applies to as one blob, the model has to guess where your command ends and y
Stop pasting paragraphs onto slides. This prompt rewrites your text into slide-ready lines plus speaker notes
The most common deck mistake is dumping full sentences onto the slide and then reading them out loud. This prompt splits the load: short lines on the slide, the real detail in the notes where it belongs. ``` Rewrite the text below for a presentation. Split every point into two layers. Text: [PASTE] For each slide: - ON SLIDE: a short headline (a claim) and up to 3 bullets, each
Belge Parçalama Karar Ağacı
Belge türüne göre açıklanabilir parçalama stratejisi üretir.
Belge RAG Parçalama Deneyi
RAG parçalama kararını tahmin yerine kontrollü deneyle verir.
Graph RAG Uygunluk Analizi
Bilgi grafiği yatırımını somut soru türleri ve maliyetle gerekçelendirir.
RAG Getirme Katmanı Teşhisi
RAG hatasını üretimden önce getirme zincirinin doğru aşamasında bulur.
RAG Hata Ayıklama Otopsisi
RAG yanıtlarındaki sorunun kaynağını kanıta dayalı biçimde ayırır.