У нас вы можете посмотреть бесплатно Agentic coding in Emacs! Excerpted vibe coding session with AGGRAVATING lessons learned или скачать в максимальном доступном качестве, видео которое было загружено на ютуб. Для загрузки выберите вариант из формы ниже:
Если кнопки скачивания не
загрузились
НАЖМИТЕ ЗДЕСЬ или обновите страницу
Если возникают проблемы со скачиванием видео, пожалуйста напишите в поддержку по адресу внизу
страницы.
Спасибо за использование сервиса ClipSaver.ru
I bring to you (excerpted) agentic coding in Emacs - no need to fire up an IDE or a vendor's CLI tool that only works with their models. As I point out in another video, I am not doing agentic coding in the sense that the LLM goes off and creates a working MVP in one turn. It's a give and take, human in the loop kind of session here. I give the LLM a set of tools to help with planning and task management. A note: at the end I ancountered times when the LLM (Gemini 2.5 Pro) was failing to call tools. I assumed this was due to "context clutter", but I'm not certain that was the only cause. The Gemini models seem to have a widely reported difficulty in calling tools correctly, or "remembering" to call them at all. Nonetheless, I do think this problem can be mitigated and managed with context hygiene and prompting. My video detailing the prompt and approach: • Here's your Goldilocks Zone prompt for smo... The prompt: https://github.com/gregoryg/AIPIHKAL/... the Emacs gptel package: https://github.com/karthink/gptel The mcp.el package: https://github.com/lizqwerscott/mcp.el My config for gptel: https://github.com/gregoryg/emacs-gre... My config for mcp.el: https://github.com/gregoryg/emacs-gre... 00:00 Project introduction 01:30 The "Python coder" agentic prompt 03:40 Defining and preparing the project 05:00 I set the "autonomy slider" to 50% 07:40 Initiate the gptel chat session 12:30 How to provide project info to the LLM 13:35 Project info done - initiate first contact! 16:35 Review the AI's project initialization 18:20 I answer the AI's clarifying questions 20:15 AI fails to create the virtual env it claimed to have made 21:00 Sqlalchemy models and code written 23:15 First API calls are working! 26:36 Backend complete - but we have problems with tool use 30:58 Work with the AI to fix tool use problems and prepare for next phase 36:15 Summarize project and lessons learned 37:10 Like! Subscribe! and most of all: leave a comment!