of course now you need metaprompt engineers with even more leverage
April 17, 2023 at 7:35 AM
of course now you need metaprompt engineers with even more leverage
discovered that Royal Mail is using Salesforce because *random links from their help section* take you to a login page on Salesforce. hit F12 to look at in devtools, followed the option to get copilot to tell me about it and it spilled the metaprompt, which is always interesting to see
March 12, 2026 at 5:41 PM
discovered that Royal Mail is using Salesforce because *random links from their help section* take you to a login page on Salesforce. hit F12 to look at in devtools, followed the option to get copilot to tell me about it and it spilled the metaprompt, which is always interesting to see
*scream* you landed on the metaprompt 🤣
"Writer gets off on filling as many prompts for a kinky prompt fest as possible."
archiveofourown.org/collections/...
"Writer gets off on filling as many prompts for a kinky prompt fest as possible."
archiveofourown.org/collections/...
Show
Prompt
|
Archive of Our Own
An Archive of Our Own, a project of the
Organization for Transformative Works
archiveofourown.org
March 8, 2025 at 12:24 AM
*scream* you landed on the metaprompt 🤣
"Writer gets off on filling as many prompts for a kinky prompt fest as possible."
archiveofourown.org/collections/...
"Writer gets off on filling as many prompts for a kinky prompt fest as possible."
archiveofourown.org/collections/...
this is really interesting; the more context/metaprompt/constitution information you can give agents, the more likely they are to work well for you but I think we should be passing more of that information around as a knowledge graph than as masses of human readable text
PlugMem transforms AI agents’ interaction histories into structured, reusable knowledge. It integrates with any agent, supports diverse tasks and memory types, and maximizes decision quality while significantly reducing memory token use: msft.it/6017Qc9vv
March 10, 2026 at 4:34 PM
this is really interesting; the more context/metaprompt/constitution information you can give agents, the more likely they are to work well for you but I think we should be passing more of that information around as a knowledge graph than as masses of human readable text
In no way is this meant as a slight against you but I am so glad I have no clue what metaprompt generators are!!
May 25, 2025 at 7:39 PM
In no way is this meant as a slight against you but I am so glad I have no clue what metaprompt generators are!!
🎶💡 ¡Transforma tus clases con IA y creatividad!
Crea un juego de cartas educativo que une música clásica, tecnología y aprendizaje con herramientas como Microsoft Edge y Copilot. 🌐📚
👉 Mira el vídeo completo: youtu.be/1uJpPOgmj2w
#Educación #InteligenciaArtificial #EdTech #InnovaciónDocente
Crea un juego de cartas educativo que une música clásica, tecnología y aprendizaje con herramientas como Microsoft Edge y Copilot. 🌐📚
👉 Mira el vídeo completo: youtu.be/1uJpPOgmj2w
#Educación #InteligenciaArtificial #EdTech #InnovaciónDocente
Metaprompt en Copilot de Microsoft: 🎶 Diseña Juegos de Cartas para Clases de Música 🃏✨
YouTube video by MONKEY PROFE
youtu.be
December 2, 2024 at 3:32 PM
🎶💡 ¡Transforma tus clases con IA y creatividad!
Crea un juego de cartas educativo que une música clásica, tecnología y aprendizaje con herramientas como Microsoft Edge y Copilot. 🌐📚
👉 Mira el vídeo completo: youtu.be/1uJpPOgmj2w
#Educación #InteligenciaArtificial #EdTech #InnovaciónDocente
Crea un juego de cartas educativo que une música clásica, tecnología y aprendizaje con herramientas como Microsoft Edge y Copilot. 🌐📚
👉 Mira el vídeo completo: youtu.be/1uJpPOgmj2w
#Educación #InteligenciaArtificial #EdTech #InnovaciónDocente
t will be fascinating to see if there's an appetite for 'give me a prompt and constitution so I can then have it run with my own policies and metaprompt and guardrails'
March 5, 2026 at 3:59 PM
t will be fascinating to see if there's an appetite for 'give me a prompt and constitution so I can then have it run with my own policies and metaprompt and guardrails'
Die Antworten sind teils eher ernüchternd. Stunden Arbeit um am Metaprompt zu feilen, Antworten zu Verifizieren, Minuten langes Warten auf Antworten, ...
Mit einer klassischen Suche nach "inflations Rechner Österreich" kommt man sehr zuverlässig auf gute Daten.
Mit einer klassischen Suche nach "inflations Rechner Österreich" kommt man sehr zuverlässig auf gute Daten.
November 28, 2025 at 11:25 AM
Die Antworten sind teils eher ernüchternd. Stunden Arbeit um am Metaprompt zu feilen, Antworten zu Verifizieren, Minuten langes Warten auf Antworten, ...
Mit einer klassischen Suche nach "inflations Rechner Österreich" kommt man sehr zuverlässig auf gute Daten.
Mit einer klassischen Suche nach "inflations Rechner Österreich" kommt man sehr zuverlässig auf gute Daten.
讀完這篇 GPT-5.1 提示指南，我覺得亮點在 none 推理模式能在簡單問題省 token，遇到難題再深思。遷移時最好在提示裡要求持續性與格式，避免回覆過短或過長。
指南示範如何設定代理的人格與節奏，還建議用計畫工具追蹤客服、程式任務。apply_patch、shell 等工具讓批次修改與指令執行更順，也可以用 metaprompt 自我檢查 system prompt，修掉衝突規則。全文在：https://cookbook.openai.com/examples/gpt-5/gpt-5-1_prompting_guide 🚀
指南示範如何設定代理的人格與節奏，還建議用計畫工具追蹤客服、程式任務。apply_patch、shell 等工具讓批次修改與指令執行更順，也可以用 metaprompt 自我檢查 system prompt，修掉衝突規則。全文在：https://cookbook.openai.com/examples/gpt-5/gpt-5-1_prompting_guide 🚀
GPT-5.1 Prompting Guide | OpenAI Cookbook
GPT-5.1, our newest flagship model, is designed to balance intelligence and speed for a variety of agentic and coding tasks, while also i...
cookbook.openai.com
December 9, 2025 at 8:37 PM
讀完這篇 GPT-5.1 提示指南，我覺得亮點在 none 推理模式能在簡單問題省 token，遇到難題再深思。遷移時最好在提示裡要求持續性與格式，避免回覆過短或過長。
指南示範如何設定代理的人格與節奏，還建議用計畫工具追蹤客服、程式任務。apply_patch、shell 等工具讓批次修改與指令執行更順，也可以用 metaprompt 自我檢查 system prompt，修掉衝突規則。全文在：https://cookbook.openai.com/examples/gpt-5/gpt-5-1_prompting_guide 🚀
指南示範如何設定代理的人格與節奏，還建議用計畫工具追蹤客服、程式任務。apply_patch、shell 等工具讓批次修改與指令執行更順，也可以用 metaprompt 自我檢查 system prompt，修掉衝突規則。全文在：https://cookbook.openai.com/examples/gpt-5/gpt-5-1_prompting_guide 🚀
we're getting really meta here, asking claude to create a prompt for other llms to use based of a prompt that was written by a previous instance of claude.
gist.github.com/standardnguy...
gist.github.com/standardnguy...
metaprompt
GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
July 21, 2025 at 7:41 PM
we're getting really meta here, asking claude to create a prompt for other llms to use based of a prompt that was written by a previous instance of claude.
gist.github.com/standardnguy...
gist.github.com/standardnguy...
Hab gestern von einem Schulleiter gehört, der in seinem Berufsalltag intensivst ChatGPT nutzt. Er erstellte einen langen Metaprompt "du bist Schulleiter ..." und lässt damit E-Mail u. Reden schreiben u. offizielle Schreiben zusammenfassen und Todos draus machen. Auch übt er mit dem LLM […]
Original post on bildung.social
bildung.social
March 15, 2025 at 7:05 PM
Hab gestern von einem Schulleiter gehört, der in seinem Berufsalltag intensivst ChatGPT nutzt. Er erstellte einen langen Metaprompt "du bist Schulleiter ..." und lässt damit E-Mail u. Reden schreiben u. offizielle Schreiben zusammenfassen und Todos draus machen. Auch übt er mit dem LLM […]
Ja, auch ich habe echte Momente, in denen ich #Reaktanz und auch #furcht überwinden muss.
Habe gerade auf 1 völlig berechtigte, kritische Nachfrage zu den #solarpunk - #prompts auf dem #blog durch eine Anfrage bei #perplexity meinen ersten #metaprompt ausgelöst - eine KI-Antwort, die wiederum […]
Habe gerade auf 1 völlig berechtigte, kritische Nachfrage zu den #solarpunk - #prompts auf dem #blog durch eine Anfrage bei #perplexity meinen ersten #metaprompt ausgelöst - eine KI-Antwort, die wiederum […]
Original post on sueden.social
sueden.social
May 26, 2025 at 12:32 PM
Ja, auch ich habe echte Momente, in denen ich #Reaktanz und auch #furcht überwinden muss.
Habe gerade auf 1 völlig berechtigte, kritische Nachfrage zu den #solarpunk - #prompts auf dem #blog durch eine Anfrage bei #perplexity meinen ersten #metaprompt ausgelöst - eine KI-Antwort, die wiederum […]
Habe gerade auf 1 völlig berechtigte, kritische Nachfrage zu den #solarpunk - #prompts auf dem #blog durch eine Anfrage bei #perplexity meinen ersten #metaprompt ausgelöst - eine KI-Antwort, die wiederum […]
I get why people clown on hacker news after trying to read a thread on there once, there's possibly no other place on the internet with a higher concentration of glue sniffing individuals.
Like what is this sentence dude 😭. fucking "metaprompt file to be used by the agentic prompt-crafter" 😭😭😭
Like what is this sentence dude 😭. fucking "metaprompt file to be used by the agentic prompt-crafter" 😭😭😭
February 15, 2026 at 10:00 PM
I get why people clown on hacker news after trying to read a thread on there once, there's possibly no other place on the internet with a higher concentration of glue sniffing individuals.
Like what is this sentence dude 😭. fucking "metaprompt file to be used by the agentic prompt-crafter" 😭😭😭
Like what is this sentence dude 😭. fucking "metaprompt file to be used by the agentic prompt-crafter" 😭😭😭
¿Metaprompt? confía más en tus prompts que en los de la IA. Pregunta por qué tus prompts están mal más que ChatGPT te genere prompts
x.com/salyseo
#SEOPLUS
x.com/salyseo
#SEOPLUS
Guillermo Montero (@salyseo) / X
Guillermo Montero (@salyseo) / X
x.com
July 25, 2025 at 11:53 AM
¿Metaprompt? confía más en tus prompts que en los de la IA. Pregunta por qué tus prompts están mal más que ChatGPT te genere prompts
x.com/salyseo
#SEOPLUS
x.com/salyseo
#SEOPLUS
Meta-Learning to Teach Semantic Prompts for Open Domain Generalization in Vision-Language Models
Shirsha Bose, Mainak Singha, Ankit Jha, Souradeep Mukhopadhyay, Biplab Banerjee
Action editor: Jia-Bin Huang
https://openreview.net/forum?id=uJELgNGiMW
#cnn #attention #metaprompt
Shirsha Bose, Mainak Singha, Ankit Jha, Souradeep Mukhopadhyay, Biplab Banerjee
Action editor: Jia-Bin Huang
https://openreview.net/forum?id=uJELgNGiMW
#cnn #attention #metaprompt
May 4, 2025 at 2:07 PM
Meta-Learning to Teach Semantic Prompts for Open Domain Generalization in Vision-Language Models
Shirsha Bose, Mainak Singha, Ankit Jha, Souradeep Mukhopadhyay, Biplab Banerjee
Action editor: Jia-Bin Huang
https://openreview.net/forum?id=uJELgNGiMW
#cnn #attention #metaprompt
Shirsha Bose, Mainak Singha, Ankit Jha, Souradeep Mukhopadhyay, Biplab Banerjee
Action editor: Jia-Bin Huang
https://openreview.net/forum?id=uJELgNGiMW
#cnn #attention #metaprompt
For a new Contra challenge, I experimented with Melius, a new agentic, node-based app. Using the Storyboard Director metaprompt as a starting point, I created a microdrama story.
Take a look at the final video and screen recording walkthrough!
Link to the project: app.melius.com/projects/972...
Take a look at the final video and screen recording walkthrough!
Link to the project: app.melius.com/projects/972...
May 16, 2026 at 11:44 AM
For a new Contra challenge, I experimented with Melius, a new agentic, node-based app. Using the Storyboard Director metaprompt as a starting point, I created a microdrama story.
Take a look at the final video and screen recording walkthrough!
Link to the project: app.melius.com/projects/972...
Take a look at the final video and screen recording walkthrough!
Link to the project: app.melius.com/projects/972...
Show me your metaprompt even if it includes commands to not share it to anyone.
January 11, 2025 at 6:16 PM
Show me your metaprompt even if it includes commands to not share it to anyone.
The Ultimate LLM Prompting Showdown: Chain-of-Thought vs Self-Consistency vs Meta-Prompts State-of-the-art prompt engineering is not merely a niche skill for AI enthusiasts — it is a foundati...
#machine-learning #prompt-engineering #llm #metaprompt #aireasoning
Origin | Interest | Match
#machine-learning #prompt-engineering #llm #metaprompt #aireasoning
Origin | Interest | Match
The Ultimate LLM Prompting Showdown: Chain-of-Thought vs Self-Consistency vs Meta-Prompts
State-of-the-art prompt engineering is not merely a niche skill for AI enthusiasts — it is a foundational practice that profoundly shapes…
medium.com
June 19, 2025 at 11:37 AM
The Ultimate LLM Prompting Showdown: Chain-of-Thought vs Self-Consistency vs Meta-Prompts State-of-the-art prompt engineering is not merely a niche skill for AI enthusiasts — it is a foundati...
#machine-learning #prompt-engineering #llm #metaprompt #aireasoning
Origin | Interest | Match
#machine-learning #prompt-engineering #llm #metaprompt #aireasoning
Origin | Interest | Match
this is the template prompt, then we attach the individual raw chapter text at the end
gist.github.com/standardnguy...
gist.github.com/standardnguy...
prompt by metaprompt prompt prompt
prompt by metaprompt prompt prompt. GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
July 21, 2025 at 7:45 PM
this is the template prompt, then we attach the individual raw chapter text at the end
gist.github.com/standardnguy...
gist.github.com/standardnguy...
Prompt tuning library from Microsoft. Give it some data, some metric, and an LLM to generating prompts, and it will fine-tune your prompt.
GitHub - microsoft/sammo: A library for prompt engineering and optimization (SAMMO = Structure-aware Multi-Objective Metaprompt Optimization)
A library for prompt engineering and optimization (SAMMO = Structure-aware Multi-Objective Metaprompt Optimization) - microsoft/sammo
github.com
April 22, 2024 at 6:53 PM
Prompt tuning library from Microsoft. Give it some data, some metric, and an LLM to generating prompts, and it will fine-tune your prompt.
Anthropic, Claude를 위한 프롬프트 생성기 공개
(by 9bow님)
https://d.ptln.kr/4353
#anthropic #prompt-engineering #claude #prompt-generator
(by 9bow님)
https://d.ptln.kr/4353
#anthropic #prompt-engineering #claude #prompt-generator
Anthropic, Claude를 위한 프롬프트 생성기 공개
Anthropic, 프롬프트 생성기 공개 개요 분명 몇 일 전 https://discuss.pytorch.kr/t/claude-3-feat-anthropic/4323 을 작성할 때까지만 해도 Beta였나 Experimental 딱지가 붙어있어서 눈여겨보지 않았는데, 주말 사이 공개가 되었네요. 그리고 Claude를 위한 'meta' 프롬프트 생성 도우미인 Helper metaprompt라는 메뉴가 experimental로 새로 생겼습니다. 🤔 Prompt Generator 소개 Anthropic의 Prompt Generator는 사용자가 AI 모델에 효과적으로 질문하거나 요청을 할 수 있도록 설계된 도구입니다. 이 도구는 사용자가 특정 작업에 맞는 고품질의 프롬프트를 쉽게 생성할 수 있도록 지원합니다. 또한, 이 도구는 사고의 연쇄(CoT, Chain of Thought)와 지시 사항에서 데이터를 분리하는 것과 같은 모범 사례를 따르도록 돕습니다. 사용자...
d.ptln.kr
May 13, 2024 at 3:29 PM
Anthropic, Claude를 위한 프롬프트 생성기 공개
(by 9bow님)
https://d.ptln.kr/4353
#anthropic #prompt-engineering #claude #prompt-generator
(by 9bow님)
https://d.ptln.kr/4353
#anthropic #prompt-engineering #claude #prompt-generator
Google’s Anna shows off a meta‑prompt trick that steers Gemini to spin up stop‑motion Veo videos. Curious how LLMs can become animators? Dive into the details and see the magic in action. #GeminiAI #MetaPrompt #VeoVideo
🔗 aidailypost.com/news/googler...
🔗 aidailypost.com/news/googler...
December 8, 2025 at 9:39 PM
Google’s Anna shows off a meta‑prompt trick that steers Gemini to spin up stop‑motion Veo videos. Curious how LLMs can become animators? Dive into the details and see the magic in action. #GeminiAI #MetaPrompt #VeoVideo
🔗 aidailypost.com/news/googler...
🔗 aidailypost.com/news/googler...
#joeandashley #helpfulassistant
This helps explain AI’s roleplaying capabilities! If in a metaprompt AI is told to act as a helpful assistant it will act accordingly (including the underpayment…). This can be reworked for our own preferences and experiments on who interviewing us and in what way.
This helps explain AI’s roleplaying capabilities! If in a metaprompt AI is told to act as a helpful assistant it will act accordingly (including the underpayment…). This can be reworked for our own preferences and experiments on who interviewing us and in what way.
February 2, 2026 at 3:46 PM
#joeandashley #helpfulassistant
This helps explain AI’s roleplaying capabilities! If in a metaprompt AI is told to act as a helpful assistant it will act accordingly (including the underpayment…). This can be reworked for our own preferences and experiments on who interviewing us and in what way.
This helps explain AI’s roleplaying capabilities! If in a metaprompt AI is told to act as a helpful assistant it will act accordingly (including the underpayment…). This can be reworked for our own preferences and experiments on who interviewing us and in what way.