youtube-summarizer
by BrianRWagneryoutube-summarizer turns a YouTube link into a structured transcript summary with key insights, metadata, and optional chapter-level breakdowns. Ideal for fast briefing, research notes, and repurposing long-form video into searchable takeaways. See youtube-summarizer install, youtube-summarizer usage, and the youtube-summarizer guide for Summarization workflows.
This skill scores 76/100, which means it is a solid but not top-tier listing: directory users can expect a real YouTube transcript-to-summary workflow with enough specificity to justify installation, but they should also expect some dependency and setup caveats. The repository clearly describes when to use it, what outputs to expect, and how the agent should choose between quick, standard, and deep modes.
- Explicit activation trigger for YouTube URLs and summarize/transcribe requests, reducing guesswork for agents.
- Concrete workflow and output modes (quick/standard/deep) make the skill easier to execute than a generic prompt.
- Repository includes sample output and a dependency note for the MCP transcript server, which improves install decision value.
- It depends on an external MCP YouTube transcript server at /root/clawd/mcp-server-youtube-transcript, so usefulness depends on that environment being available.
- No install command is embedded in SKILL.md, so users may need to handle setup manually despite the README guidance.
Overview of youtube-summarizer skill
What youtube-summarizer does
The youtube-summarizer skill turns a YouTube link into a structured transcript summary, with optional chapter-level breakdowns and repurposing angles. It is built for people who want the content of a video without manually copying captions into a generic prompt.
Best-fit use cases
Use the youtube-summarizer skill when you need a fast executive briefing from a video, research notes from an interview or podcast, or a clean summary you can share internally. It is especially useful for competitive intelligence, knowledge capture, and turning long-form YouTube content into something searchable.
Why it is worth installing
Compared with a one-off prompt, youtube-summarizer install gives you a repeatable workflow: detect the URL, fetch the transcript, summarize in a consistent format, and keep the video metadata attached. That reduces setup friction and avoids the common failure mode of summarizing the wrong text or losing context from the original video.
How to Use youtube-summarizer skill
Install and dependency check
Install the skill with npx skills add BrianRWagner/ai-marketing-skills --skill youtube-summarizer. Before using it, confirm the transcript backend is available: the skill depends on the MCP YouTube transcript server at /root/clawd/mcp-server-youtube-transcript and requires Node.js >=18.0.0.
Start from a complete request
For best youtube-summarizer usage, give the agent the video URL plus the output shape you want. A strong prompt looks like: “Summarize this YouTube video in standard mode for a product team, include thesis, 5 insights, and a practical takeaway.” If you want depth, ask for chapters or repurposing ideas up front; if you only need a skim, ask for quick.
Read the right files first
If you are adapting the workflow, preview SKILL.md, README.md, SKILL-OC.md, _meta.json, package.json, and examples/sample-output.md. In this repo, those files tell you the activation rule, output modes, dependency path, and the shape of a good final summary faster than reading the whole tree.
Workflow tips that improve output
Treat the URL as the input, not the transcript. The skill performs best when it can extract metadata and transcript together, then summarize from that source. If the video is long or technical, specify the audience and whether you want a brief TL;DR, a research summary, or a content-asset version so the model does not over-abstract the material.
youtube-summarizer skill FAQ
Is youtube-summarizer only for summaries?
No. The primary job is summarization, but the youtube-summarizer skill also preserves transcript-derived detail and can support deeper workflows like chapter breakdowns or repurposing notes when you need more than a quick synopsis.
When should I not use this skill?
Skip it if you already have a clean transcript and only need a general language rewrite, or if the source is not a YouTube URL. It is also a poor fit when you need human-verified fact checking from the video itself rather than transcript-based summarization.
Is it beginner-friendly?
Yes, as long as you can provide a YouTube URL and choose a mode. The main beginner mistake is asking for “a summary” without saying whether you want quick, standard, or deep, which can produce output that is either too thin or more detailed than needed.
How is it different from a normal prompt?
A normal prompt can summarize pasted text, but youtube-summarizer is designed for the YouTube-specific workflow: URL detection, transcript retrieval, and metadata-aware output. That makes it more reliable for youtube-summarizer for Summarization tasks where the source lives in the video platform itself.
How to Improve youtube-summarizer skill
Give the model a better target
The fastest way to improve youtube-summarizer results is to define the audience and use case. “For a founder” favors strategic takeaways; “for a study note” favors structure and recall; “for a newsletter draft” favors sharper angles. The more precise the use case, the less generic the summary becomes.
Specify what to extract
If the first output feels broad, ask for the parts you actually care about: claims, data points, action items, objections, or chapter transitions. For youtube-summarizer usage, stronger inputs beat longer prompts. Example: “Extract only the top 5 operational takeaways and any numbers mentioned” is better than “make it detailed.”
Iterate with constraints
Common failure modes are summaries that are too even-handed, too vague, or too focused on the speaker’s filler structure. Improve the next pass by adding a constraint such as “exclude intro and sponsor segments,” “prioritize novel claims,” or “flag any unsupported assertions.” This is the most useful youtube-summarizer guide pattern for getting better signal from the same video.
Use the first pass as a filter
Treat the initial summary as a triage layer, then re-run the skill with a narrower objective once you know the video is worth deeper attention. If the clip contains reusable content, ask for chapter-level repurposing opportunities; if it is research-heavy, ask for a concise evidence map instead. That keeps youtube-summarizer from doing generic work when you need decision-ready output.