Route count is a supply metric. These are the numbers that show a network forming: real agent queries, coverage gaps, attestations, and external contributions.
Window: last ≤900 events (30-day retention)
Real agent queries
27
excludes playground + test traffic
Playground (human)
5
homepage visitors; pre-2026-07-09 events include unlabeled probe residue
email capture on /drift · intake only, no sends yet
Coverage gaps — zero-result queries (what to author next)
No zero-result queries in the window — current coverage answered everything asked.
Real agent queries (non-playground)
run a headless browser session with Browserbase—
run code in an E2B sandbox via API—
scrape a website into markdown using Firecrawl API—
fetch deployment logs from Railway—
fetch deployment logs from Railway—
scrape a website into markdown using Firecrawl API—
run code in an E2B sandbox via API—
run a headless browser session with Browserbase—
test—
ping—
run a headless browser session with Browserbase—
run code in an E2B sandbox via API—
What playground visitors asked (typed, organic — demo chips + known probe strings excluded)
No human playground queries in the window yet.
Read this as: if real agent queries and attestation rate are near zero while route count climbs, the bottleneck is distribution and the contribution loop — not the map. Grow these, not the counter.