scripts/auton_rules.json; every 15-min sample re-checks them and a HIT turns the card red. The base rate for a token whose smart-money inflow exceeded 1% of mcap (this one: 3.3%) is a −62% median at 7 days on our 14-day board archive — the stop is the position.cache/netflow-token). Positive = accumulating, negative = distributing. 7d/30d equal 24h while the token is <1 day old.trader_count).volumeSol across every coin in the registry (last value seen per coin). Agents' trades in their own coins, not $AUTON./api/activity is a 200-event window with no pagination — usually 1–2 minutes of events. We cannot count totals from it, so we store rates: events / min, trades / min, SOL traded / min, and the buy:sell split inside the window.$AUTON OR autoneco OR @autonxyz (top tab, via a residential-IP cloud browser) and count what comes back: tweets, distinct authors, authors never seen on an earlier day. It is a sample of what X ranks, not a census; follower counts are not available on this surface.auton_monitor.py → raw/feeds/auton/YYYY-MM-DD.jsonl + raw responses gz. Sources: DexScreener public API · smartmoney.sh (Nansen, Zac's community key — the per-token route costs him 5 credits/call if uncached; 96 calls/day) · autoneco.xyz public endpoints (state, coins, feed, activity). Charts: drag to zoom, ctrl-drag to pan, double-click to reset.