Сравнение Claude Code, Gemini и Codex — тесты на контент, анализ и изображения
The speaker is comparing AI coding assistants — Claude Code, Gemini, and Codex — by testing their ability to create image-generation agents. Each tool is being evaluated on how well it writes prompts and structures sub-agents for text-to-image and image-to-image tasks.
Summary
The transcript covers a live comparison test of three AI coding tools: Claude Code, Gemini (referred to as 'Rin'), and Codex. The speaker is evaluating how each tool handles the creation of an image-generation agent, specifically one that can distinguish between text-to-image and image-to-image workflows.
The speaker notes that one of the tools (likely Codex or Gemini) has already written a working prompt and registered a sub-agent in its framework, with a detailed English-language prompt that the speaker personally prefers. The speaker appreciates that this tool clearly divided the use cases — separating text-to-image and image-to-image modes — and wrote specific instructions for each.
At the time of recording, Claude Code is still processing and has been working longer than the others. The speaker is waiting to see whether Claude Code will also produce its own sub-agent. The overall tone is observational and comparative, with the speaker reserving final judgment until all three tools complete their tasks.
Key Insights
- The speaker prefers detailed English-language prompts, arguing that tools cope better with tasks when the prompt is more thorough and specific.
- One of the tools (likely Codex or Gemini) distinguished between text-to-image and image-to-image modes and wrote separate instructions for each use case, which the speaker highlighted as a positive structural decision.
- The speaker notes that a sub-agent was successfully registered in 'Rin' (Gemini) using a detailed specifier, suggesting the tool handled agent architecture correctly.
- Claude Code was taking longer to complete the agent creation task compared to the other tools at the time of the test.
- By approximately the 1:02 mark, all three tools had each created an agent, though the speaker was still waiting on Claude Code to confirm whether it had produced its own sub-agent.
Topics
Transcript
[0:01] He said that it was okay that he had added a new sapagent to the image creator and had written a working prompt for it. How did you register an agent in Rin? Created a detailed specifier. Okay, let's now enter the codex as well . Here is the code of clow. And here he said that he has special agents here in the special agent. So, the code has already been written here, look, the promt is more detailed in [0:31] English. I somehow like it better, because here, when there is a detailed prompt, it copes better. He described the tasks, that is, tek to image, that we use tek to image. When we image to image,…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Игорь Зуевич
Экономия токенов и контекстного окна: настройка памяти нейросети – Compact сжатие контекстного окна
The transcript explains the 'Compact' command in AI agent sessions, which compresses the entire conversation history to reduce context window usage without switching sessions. It also covers related commands for full history clearing and automatic triggers for compact execution. The speaker recommends configuring automatic notifications or auto-execution when the context window reaches a certain threshold.
HR-агент на Claude Code: создаёт других сотрудников сам по запросу
The video demonstrates setting up a Telegram HR bot using Claude Code, walking through the process of creating a bot via BotFather, obtaining a Telegram user ID, and letting the code automatically generate the HR agent. The presenter shows how the system initializes itself by installing dependencies and launching the bot with minimal manual configuration.
Живые скиллы для Claude Code, Gemini и Codex | Как настроить super skills для ИИ агентов
The video explains how to create 'living' or 'super' skills for AI agents like Claude Code, Gemini, and Codex that self-update based on user feedback. The presenter demonstrates how these skills collect feedback after each task, store it in log files, and periodically incorporate that feedback into the skill's core instructions. This approach reduces token consumption and ensures skills continuously improve over time.
Планировщик для ИИ агента: 50 идей и готовые файлы для Claude, Codex и Gemini
The speaker describes their use of a scheduling/routing tool called Routins, which they compare to N8N, to automate daily delivery of curated YouTube videos on topics like Gemini and Claude. The system runs autonomously, collecting videos and generating concise summaries of their content without manual intervention.
Экономия токенов и контекстного окна: настройка памяти нейросети Newchat без потери данных
The transcript explains the Newchat command, which opens a new session while pulling data from previous sessions. It works in conjunction with the checkpoint command, which can be configured to trigger automatically when Newchat is launched. The speaker recommends implementing these commands through skills rather than MD slots for a more professional workflow.