A content-research agent that keeps the source clean
Use the website-to-markdown skill or recipe when the source is a page or site, and the YouTube transcript recipe when the source is a video. Send the output to Notion or a markdown store, keep the source URL and do not treat the extraction as a full-site crawl.
RAG quality starts with extraction quality. Navigation, cookie banners, duplicate blocks and truncated transcripts all become context noise. The files here produce a bounded markdown result and keep the source identifiable.
Method
Choose the source type: a website, a single page or a YouTube video.
Pick the markdown recipe or the transcript recipe. Check the input schema before the run.
Send the output to Notion or a file store. Preserve the source URL and the extraction date.
Clean the markdown before chunking. Remove navigation, repeated headers and empty sections.
Recipes for this job
YouTube transcripts to Notion
Turn public YouTube transcripts into Notion pages for a research library or RAG corpus.
YouTubeNotionGoogle News monitoring to Slack
Post the last day's Google News articles for a topic to Slack, with headline, publisher and link.
Google NewsSlackWebsite to markdown to Notion
Crawl a documentation site into markdown on a schedule and file each page in Notion, again whenever a page's…
Any public websiteNotionProduct Hunt launches to Slack
Post each day's Product Hunt launches to Slack with tagline, votes and link, so your team sees new tools in…
Product HuntSlackGitHub trending repos to Discord
Share GitHub's trending repositories in a Discord channel on a schedule, with language, stars and description.
GitHubDiscordGoogle Trends to Google Sheets
Save Google Trends interest over time and related queries for your terms in a Google Sheet to spot rising…
Google TrendsGoogle SheetsGoogle Scholar papers to Notion
File Google Scholar search results in a Notion database with authors, year, citations and link, as a reading…
Google ScholarNotionCrunchbase funding news to Slack
Post new Crunchbase News funding stories to Slack as a daily deal-flow feed.
Crunchbase NewsSlack
MCP servers for this job
- YouTube MCP server17 Actors
Skills for this job
Guardrails
- A page extraction is not a full crawl. A site-wide job needs a crawler and a crawl policy.
- A transcript can be auto-generated and may contain errors.
- Do not republish third-party content without checking copyright and the source terms.
Questions
Can this agent crawl an entire website?
The recipe handles a bounded page or URL input. A full-site crawl needs a crawler, a scope limit and a policy for depth and rate. Do not treat one page extraction as a complete site archive.
Are YouTube transcripts always accurate?
No. Auto-generated captions can contain errors. Treat a transcript as source material and keep the video URL for verification.
Where should the markdown go?
Notion works for a reviewable knowledge base; a file store works for a RAG pipeline. Either way, keep the source URL and the extraction date with the document.