EmDash Perf Monitor — HTTP API for agents (humans: use the dashboard below). Same origin, JSON responses. Query these directly instead of scraping the rendered charts. GET /api/config — { sites:[{id,label,targetUrl}], defaultSite, routes, regions:[{id,label}] }. Authoritative list of valid site/route/region ids and their display labels. GET /api/summary?site=ID — latest sample per route+region plus rolling 7-day averages: { site, latest:[…], medians:[{route,region,median_cold,median_warm,count}], config }. GET /api/chart?route=PATH®ion=ID&site=ID&since=ISO8601[&bucket=day][&limit=N] — time series for one route+region. Pass bucket=day for ranges beyond ~2 days: it returns one true-median point per UTC day, so the window isn't truncated. Without bucket=day you get raw samples newest-first capped at limit (default 200) — at 48 samples/day that only spans ~7-10 days regardless of since. Response: { route, region, site, data:[{timestamp,coldTtfbMs,warmTtfbMs,p95TtfbMs}], deployMarkers:[…] }. GET /api/results?route=®ion=&source=deploy|cron|manual&site=&since=ISO8601&limit=N — raw rows newest-first; all filters optional; omit site to span every site. → { results:[…] }. POST /api/trigger — JSON body, all optional { note, sha, prNumber, ephemeral, site }. Runs an on-demand measurement (source=manual) and returns its summary.