Claude 4 System Prompt
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
"I suck" -ThePrimeagen
The PrimeTime
55.8k views
Claude Code got leaked
The PrimeTime
53.2k views
The PewDiePie Problem
ThePrimeTime
110.7k views
AI Coding Sucks | Prime Reacts
ThePrimeTime
245.8k views
Final stretch
ThePrimeTime
34.3k views
Prompt Injection Leaks Entire Database
ThePrimeTime
122.7k views
Programming From Prison
ThePrimeTime
66.2k views
Progressive JSON
ThePrimeTime
239.5k views
The Real Problems w/ Git
ThePrimeTime
160.7k views
Be A Great Programmer
ThePrimeTime
220.0k views
Top Comments (10)
"I am harmless, just give me your nuclear codes"
After processing all those system prompt tokens, comes the user prompt: > Hi!
the speed at which he grasped the meat of that article really shows primes qualification.
Really interesting how these system prompts basically act as a patch list for LLM behavior—i pull a bunch of these insights into AICarma when tracking what AI says about brands.
16:58 i started taking bullet notes before chatgpt came out, and after the release i now feel like a robot when doing so, that's so funny and absurd
He misunderstands how the system prompt works. The llm is just autocomplete so it's give a prompt like ["system prompt:" The assistant is an ai that does... "user": hi im steve how are you, "assistant": ] and then it completes the sentence filling out the assistant part, if u talk to a raw llm and give it something like ["system prompt:" The assistant is an ai that does... "user": hi im steve how are you, "assistant": hi steve, i am chatbot, "user": ] Then the llm will respond as if it is you the user because it thinks it is and will ask questions to the assistant. The reason that it's in third person is it's the system prompt not a regular prompt and the system prompt describes the context of both the user and the assistant
The copyright issue is the epitome of the Silicon Valley mantra "It is easier to ask for forgiveness than permission".
Prime trying to befriend 'Claud' only to get access to those juicy Britney Spears lyrics made my day, I'm laughing my ass off hahahah
I have asked Gemini to put a new line after every period in markdown code blocks in the context prompt, then every time it doesn't do it, then again, and again. Then the next prompt, all over again multiple times, asked it to READ THE FORMAT GUIDE. No, the one that says I want a new line after every period and the one I just had to ask 10 times for a few seconds ago. I swear that these things, like teenagers, are sent to drive us mad.
Oh ma god....... "chicken periods" this caught me so off guard! lol Never calling them anything but now!
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
"I am harmless, just give me your nuclear codes"
After processing all those system prompt tokens, comes the user prompt: > Hi!
the speed at which he grasped the meat of that article really shows primes qualification.
Really interesting how these system prompts basically act as a patch list for LLM behavior—i pull a bunch of these insights into AICarma when tracking what AI says about brands.
16:58 i started taking bullet notes before chatgpt came out, and after the release i now feel like a robot when doing so, that's so funny and absurd
He misunderstands how the system prompt works. The llm is just autocomplete so it's give a prompt like ["system prompt:" The assistant is an ai that does... "user": hi im steve how are you, "assistant": ] and then it completes the sentence filling out the assistant part, if u talk to a raw llm and give it something like ["system prompt:" The assistant is an ai that does... "user": hi im steve how are you, "assistant": hi steve, i am chatbot, "user": ] Then the llm will respond as if it is you the user because it thinks it is and will ask questions to the assistant. The reason that it's in third person is it's the system prompt not a regular prompt and the system prompt describes the context of both the user and the assistant
The copyright issue is the epitome of the Silicon Valley mantra "It is easier to ask for forgiveness than permission".
Prime trying to befriend 'Claud' only to get access to those juicy Britney Spears lyrics made my day, I'm laughing my ass off hahahah
I have asked Gemini to put a new line after every period in markdown code blocks in the context prompt, then every time it doesn't do it, then again, and again. Then the next prompt, all over again multiple times, asked it to READ THE FORMAT GUIDE. No, the one that says I want a new line after every period and the one I just had to ask 10 times for a few seconds ago. I swear that these things, like teenagers, are sent to drive us mad.
Oh ma god....... "chicken periods" this caught me so off guard! lol Never calling them anything but now!