{"slug": "how-i-built-a-self-learning-video-editing-agent-with-claude-skills", "title": "How I Built a Self-Learning Video Editing Agent With Claude Skills", "summary": "A developer built a self-learning video editing agent using Claude Skills that can automatically edit a 30-minute video in 10 minutes. The agent separates five core editing tasks into independent Skills and features a feedback loop that personalizes the system over time. 'The more I use it, the better it understands me, and the faster it becomes,' the developer said.", "body_md": "I spent a week using video editing Skills to build a video editing Agent.\n\nIt feels amazing!\n\nIt can automatically edit a 30-minute video in just 10 minutes.\n\n*Video editing Agent demo: automatically editing a 30-minute video in 10 minutes.*\n\nI often use CapCut to edit talking-head videos, but after using it for a long time, I found several problems.\n\nBecause it cannot understand the meaning, it sometimes fails to identify repeated sections. If I speak continuously for 20 or 30 minutes, editing the video myself becomes exhausting.\n\nThe automatically generated subtitles contain many incorrect words and typos.\n\nSo I used the Skills feature in Claude Code to build a video editing Agent.\n\nThe fundamental difference is simple:\n\n*CapCut vs. Agent: a fixed tool vs. an adaptive assistant.*\n\nThe key difference is:\n\n```\nCapCut = fixed tool + manual operation\nAgent  = adaptive system + automatic learning\n```\n\nI am not replacing CapCut with a better algorithm. I am replacing it with a system that can continuously improve itself.\n\nBut that is not even the most impressive part.\n\nThe most impressive part is this: the more I use it, the better it understands me, and the faster it becomes.\n\nIt only takes four steps.\n\n*Video editing Agent workflow: from the video file to the final video.*\n\nAt first, I put every function into one large Skill. I had to add instructions to distinguish between different tasks, which was very inconvenient.\n\nNow I have separated the five core video editing tasks into five independent Skills and placed them in the `.claude/skills/`\n\ndirectory. This makes the structure clearer and the tasks easier to select.\n\nWhen I enter `/v`\n\n, Claude Code automatically lists the five available Skills.\n\n*The list of five independent Skills.*\n\nI select one, and the AI runs that Skill. Simple, right? A manual task that used to take 10 minutes now only requires selecting an item from the menu.\n\nAll of the methods are written into the Skills, so I do not need to explain them again every time.\n\nThe main reason for separating them into independent Skills is that people need to inspect the output from each stage. For example, I need to check whether the review draft is correct before confirming that the Agent should perform the actual edit.\n\nThis is the design I am most proud of.\n\nAfter each task, I can give the AI feedback, and it permanently saves that feedback into the Skills.\n\n*The self-update loop: a feedback-learning system that understands you better over time.*\n\nThe key point is that the Skills gradually change from a general set of rules into a customized solution made specifically for you.\n\nUse it 10 times, and it will understand 80% of your habits. Use it 50 times, and it will fully match your requirements.\n\nThe more you use it, the more personalized the Skills become and the better they understand you. That is the power of self-updating.\n\nOpen a new folder. You can use this folder as a dedicated workspace for video editing in the future, which makes everything more convenient.\n\nAsk Claude Code:\n\n```\nHelp me download the video editing Skills:\n\n- Repository: https://github.com/Ceeon/videocut-skills\n- Destination: .claude/skills/\n- Clone them directly into this directory without creating another subfolder.\n```\n\nIt will download the Skills automatically, and the process is very quick.\n\n*The Skills being downloaded automatically.*\n\nRestart Claude Code, and you will be able to see the Skills.\n\n*All Skills displayed in Claude Code after restarting.*\n\nEnter `/v`\n\nand select `videocut:install`\n\n.\n\n*Selecting the videocut:install command.*\n\nThe AI will automatically install the dependencies and download the models, which are about 5 GB in total:\n\nThe two models have different strengths. FunASR is suitable for word-level editing, such as identifying verbal mistakes and filler words. Whisper produces better subtitles.\n\nSupported AI models can also be accessed through [WisGate](https://wisgate.ai/).\n\nEnter `/v`\n\n, select `videocut:edit-talking-head`\n\n, and then give the AI the path to your video file.\n\n*Selecting videocut:edit-talking-head and entering the video path.*\n\nThe AI will automatically:\n\n*An example of the review draft generated for a talking-head video.*\n\nIn the review draft, the pause symbol indicates a silent section that needs to be deleted. A red dot followed by a wavy line indicates a repeated section that also needs to be deleted. If you prefer another format, you can ask the AI to change it to match your requirements.\n\nThe key step is to review the draft after it has been generated.\n\nIf anything is unsatisfactory, such as wanting to keep more filler words or correcting a word that was recognized incorrectly, continue talking with the AI until the result is right.\n\nThen enter `/v`\n\n, select `videocut:self-update`\n\n, and the AI will save the adjustments from this task into the Skills.\n\nAfter you confirm the draft, it automatically performs the edit.\n\nThe key point is this: a custom dictionary can double subtitle accuracy.\n\nThe Skills include a dictionary file:\n\n```\n.claude/skills/videocut:subtitles/dictionary.txt\n```\n\nOpen the file, and you will see that it already contains some predefined terms.\n\nYou only need to add your own proper nouns, such as your company name, product name, and channel name:\n\n```\nCheng Feng's Channel\nAI Product Freedom\nMy Brand Name\n```\n\nBy adding the terms to the dictionary in advance, the AI can use it to correct recognition errors automatically when generating subtitles. The subtitles can be accurate from the first run.\n\nThen enter `/v`\n\nand select `videocut:subtitles`\n\n.\n\nThe AI will automatically:\n\nIf you later discover that a term is missing from the dictionary, you can select `videocut:self-update`\n\nat any time to add it.\n\nFour steps, and the entire process is automated.\n\nThe final result is almost the same as editing the video manually in CapCut!\n\nIf you often edit talking-head videos, I strongly recommend giving it a try. Make sure to use the self-update system, because that is how the Agent comes to understand you better over time.\n\nIf you have more complex requirements or run into problems while using it, you are welcome to join the group and discuss them.\n\nJoin the discussion on Discord: [https://discord.gg/GjqHPC4U2t](https://discord.gg/GjqHPC4U2t).", "url": "https://wpnews.pro/news/how-i-built-a-self-learning-video-editing-agent-with-claude-skills", "canonical_source": "https://dev.to/kevin_wong/how-i-built-a-self-learning-video-editing-agent-with-claude-skills-f96", "published_at": "2026-07-21 09:38:47+00:00", "updated_at": "2026-07-21 10:01:25.418651+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-tools", "developer-tools", "machine-learning"], "entities": ["Claude Code", "CapCut", "FunASR", "Whisper", "WisGate", "GitHub", "Ceeon"], "alternates": {"html": "https://wpnews.pro/news/how-i-built-a-self-learning-video-editing-agent-with-claude-skills", "markdown": "https://wpnews.pro/news/how-i-built-a-self-learning-video-editing-agent-with-claude-skills.md", "text": "https://wpnews.pro/news/how-i-built-a-self-learning-video-editing-agent-with-claude-skills.txt", "jsonld": "https://wpnews.pro/news/how-i-built-a-self-learning-video-editing-agent-with-claude-skills.jsonld"}}