Local markdown vaults reward steady timestamps. Trends MCP feeds those vaults with dated search, news, and feed snapshots so daily notes stay anchored to what the web actually looked like when the entry was written.
Raster grabs hide the query string, lose the axis window, and resist text search. A markdown table with the keyword, source, last date, and values stays diffable in git and readable in Obsidian graph views.
Keep /Trends/YYYY/ with one file per week. Inside each file, use level-two headings per platform. Under each heading, store one table per keyword batch plus a short prose line on why the batch mattered that week.
Teams that prefer Notion for collaboration can mirror the pattern described at https://www.trendsmcp.ai/notion-trend-databases-mcp. Cursor users wiring the same server inside the IDE can follow https://www.trendsmcp.ai/mcp-server-for-cursor.
If the assistant hallucinates a column header, delete the table and rerun with the tool trace visible. Vault hygiene matters more than volume; a wrong number in an evergreen note spreads false confidence during the next quarterly read.
Never paste API keys into Obsidian files that sync to a shared drive. Keep keys in the host configuration only. If a note names a customer, scrub identifiers before attaching trend pulls from public sources.
Tools for this workflow
get_trendsArchive a weekly Google Search series for a product codename inside the meeting note for that sprint.
get_trends(keyword='internal codename', source='google search', data_mode='weekly')get_growthAdd a quarterly growth snippet next to OKR commentary in a quarterly review note.
get_growth(keyword='category phrase', source='google search', percent_growth=['3M', '6M', '12M'])get_top_trendsPaste the top fifteen YouTube trending titles into a media diary note at the end of each Friday.
get_top_trends(type='YouTube Trending', limit=15)FAQ