2.5 KiB
ATLAS Terminal Baseline — April 2026
Captured on 2026-04-21 before the v2 Data Gateway migration.
Environment
- Backend: FastAPI on
127.0.0.1:8000 - Frontend: Next.js dev server on
localhost:3000 - Database: SQLite
- Browser baseline:
agent-browserChromium session
Frontend Timing
Local overview route / after network idle:
| Metric | Value |
|---|---|
| DOM interactive | 171 ms |
| DOMContentLoaded | 172 ms |
| First paint | 284 ms |
| First contentful paint | 284 ms |
| Load event end | 394 ms |
| Resource count | 26 |
Lighthouse was not run in this pass because the repo does not currently include a Lighthouse dependency or script. The browser navigation timing above is the baseline until Lighthouse is added.
API Latency Samples
Measured with curl against the live local backend. Values are seconds.
| Endpoint | Samples | Approx p50 | Approx p95 | Notes |
|---|---|---|---|---|
/api/health |
0.0014, 0.0024, 0.0010, 0.0010 | 0.0012 | 0.0024 | Stable local health check |
/api/market/indices |
0.7092, 0.8025, 0.2898, 0.3265 | 0.518 | 0.8025 | yfinance-backed and still provider-bound |
/api/market/asset-type/AAPL |
0.2486, 0.1845, 0.2014, 0.1920 | 0.2014 | 0.2486 | Lightweight ticker classification |
/api/portfolio/summary |
0.2929, 0.0019, 0.0017, 0.0018 | 0.0019 | 0.2929 | Warm cache makes repeat calls near-instant |
/api/fx/rates |
0.1722, 0.0012, 0.0010, 0.0012 | 0.0012 | 0.1722 | Warm cache makes repeat calls near-instant |
Bundle Size Baseline
Latest npm run build route output:
| Route | Size | First Load JS |
|---|---|---|
/ |
6.35 kB | 98.2 kB |
/earnings |
4.31 kB | 92 kB |
/filings |
8.24 kB | 96 kB |
/macro |
16 kB | 207 kB |
/markets |
7.03 kB | 94.8 kB |
/news |
5.43 kB | 93.2 kB |
/portfolio |
5.93 kB | 93.7 kB |
/report |
21.7 kB | 213 kB |
/research |
2.73 kB | 94.6 kB |
/screener |
3.26 kB | 91 kB |
/settings |
2.57 kB | 90.3 kB |
/technical |
6.38 kB | 94.1 kB |
/valuation |
5.1 kB | 97 kB |
Shared first-load JS: 87.7 kB.
FMP Call Baseline
Runtime FMP call volume is not yet centrally instrumented. Static inspection shows FMP access still lives behind server/services/fmp_client.py, but daily call count cannot be measured reliably until Phase 1 routes provider traffic through the Data Gateway.
Baseline tracking target for Phase 1: add provider-level counters at the gateway seam and compare FMP calls before/after each migrated endpoint.