MINDSPENT

Data and MCP

The index as open files

Read it in a spreadsheet, a notebook or an AI assistant. Data month September 2026, refreshed monthly.

Downloads

  • index.json: all 2210 companies with industries, batches and cohorts
  • board.json: the compact table the board uses
  • mindspent-yc-index-2026-09.csv
  • /data/company/<slug>.json: one file per company with 12 months of history, lens sentences, cautions, keywords and founders, for example scribd
curl -s https://mindspent.com/data/index.json | jq '.companies[0:3]'

Fields

  • visits (rounded to 3 significant figures), last3, change3, changeMonth
  • avgVisitSec, bounce, pagesPerVisit, globalRank
  • attentionMin, mindshare, rank, industryRank, batchRank, percentile (see Method)
  • searchVol, searchChange (brand searches, from DawnTrend, see Method)
  • cautions, lens, industry, category, batch, year, domain, logo

Free to use with a link back to mindspent.com. Traffic figures are panel estimates.

MCP server

Ask your assistant about YC attention.

Connect any MCP client to https://mindspent.com/mcp (Streamable HTTP, no key). Then ask things like "which 2025 batch companies doubled their visits since June" or "compare Retool and Supabase".

Claude Code

claude mcp add --transport http mindspent https://mindspent.com/mcp

Claude, Cursor and other clients

{ "mcpServers": { "mindspent": { "type": "http", "url": "https://mindspent.com/mcp" } } }

Tools (all read-only)

  • search_companiesname, industry, batch, sort
  • get_company12 months, ranks, findings, neighbours
  • top_moversrisers or fallers since June, 50K floor
  • compare_companies2 to 6 side by side
  • list_industriesshares, medians, categories
  • list_batchesevery batch, newest first
  • index_infosources, method, cautions

Same numbers as the site: panel estimates for data month September 2026, not the companies' own analytics.

Open full page