Documentation Index
Fetch the complete documentation index at: https://mintlify.com/deskiziarecords/QUIMERIA-HYPERION/llms.txt
Use this file to discover all available pages before exploring further.
QUIMERIA-HYPERION writes all pipeline activity to five rotating log files under the logs/ directory. The directory is created automatically on first startup relative to the project root. Each log file is written by a dedicated logger in backend/logger.py using Python’s RotatingFileHandler — files rotate at 10 MB and keep five backups, except raw_bars.log which rotates at 50 MB.
Log streams
| File | Content |
|---|
logs/events.log | FVGs detected, AMD state transitions, Judas swings, approved signals |
logs/veto.log | Every bar’s Ring 0 decision — trade_allowed, veto reason, and p_fused score |
logs/trades.log | Trade opens and closes with entry price, exit price, and realized P&L |
logs/session.log | Server start, data loads, live feed start/stop, module import results |
logs/raw_bars.log | Full step() result JSON for every bar (50 MB rotation limit) |
raw_bars.log grows quickly during backtests. It is intended for debugging and should not be left enabled in high-throughput production environments unless you have sufficient disk space.
Log rotation configuration
All five loggers are configured in backend/logger.py using _make_logger(). The defaults are:
| Parameter | Value |
|---|
| Max file size | 10 MB (50 MB for raw_bars.log) |
| Backup count | 5 rotated files |
| Encoding | UTF-8 |
| Timestamp format | YYYY-MM-DD HH:MM:SS UTC |
events.log.1, events.log.2, etc. The active log is always events.log.
Reading logs via the REST API
List available log files
curl http://localhost:8000/api/logs
[
"events.log",
"veto.log",
"trades.log",
"session.log",
"raw_bars.log"
]
Read a log file (last N lines)
The lines query parameter controls how many lines from the end of the file are returned. The default is 100.
curl "http://localhost:8000/api/logs/events.log?lines=200"
curl "http://localhost:8000/api/logs/veto.log?lines=50"
curl "http://localhost:8000/api/logs/trades.log?lines=100"
# Last 200 lines of events log
curl "http://localhost:8000/api/logs/events.log?lines=200"
# Last 50 veto decisions
curl "http://localhost:8000/api/logs/veto.log?lines=50"
# Trade P&L history
curl "http://localhost:8000/api/logs/trades.log?lines=100"
import httpx
r = httpx.get(
"http://localhost:8000/api/logs/events.log",
params={"lines": 200},
)
for line in r.json()["lines"]:
print(line)
Log entry formats
events.log
[2026-01-15 12:34:56 UTC] BAR#247 FVG_BULLISH low=1.10320 high=1.10580 strength=0.82
[2026-01-15 12:35:00 UTC] BAR#248 AMD_TRANSITION: ACCUMULATION → MANIPULATION
[2026-01-15 12:35:04 UTC] BAR#249 JUDAS_SWING high=1.10601 reversal_confirmed=True
[2026-01-15 12:35:08 UTC] BAR#250 SIGNAL APPROVED p_fused=0.87 confidence=0.91
veto.log
Every bar produces exactly one veto log entry with the Ring 0 decision:
[2026-01-15 12:34:56 UTC] BAR#247 VETO: trade_allowed=True decision=PROCEED p_fused=0.87
[2026-01-15 12:35:00 UTC] BAR#248 VETO: trade_allowed=False decision=HALT reason=MANDRA:ΔE<0.02
[2026-01-15 12:35:04 UTC] BAR#249 VETO: trade_allowed=False decision=HALT reason=FUSION:LAMBDA_VETO
trades.log
[2026-01-15 12:35:08 UTC] OPEN #T001 BUY EURUSD 0.10 lots @ 1.10512 SL=1.10200 TP=1.11000
[2026-01-15 13:22:41 UTC] CLOSE #T001 BUY EURUSD 0.10 lots @ 1.10998 PnL=+$48.60
session.log
[2026-01-15 10:00:01 UTC] Server startup: QUIMERIA-HYPERION v1.1
[2026-01-15 10:00:02 UTC] Modules loaded: 18/18 Import errors: 0
[2026-01-15 10:00:03 UTC] Plugins discovered: 6
[2026-01-15 10:05:12 UTC] CSV loaded: EURUSD_H1.csv bars=2340
[2026-01-15 12:34:50 UTC] Live feed started: EURUSDT 1m
TradingView dashboard panels
The frontend (frontend/index.html) displays three live log panels that stream from the same log files via the REST API:
| Panel | Source | Update frequency |
|---|
| Events panel | logs/events.log | Every bar (on WebSocket frame) |
| Veto panel | logs/veto.log | Every bar |
| Trades panel | logs/trades.log | On open/close events |
- 14 SMK sensor bars — real-time λ scores from
r['sensors']
- 6 plugin sensor bars — forensic detector scores appended to
sensors[]
- AMD state machine — current IPDA phase (Accumulate / Manipulate / Distribute / Retrace)
- Execution panel — AEGIS bridge status, veto authority indicator, and last signal ID
- TradingView Lightweight Charts v4 — candlestick chart with FVG and order block overlays
Streaming logs from the command line
For server-side monitoring without the dashboard, follow the backend log directly:
# Follow the combined backend stdout/stderr
tail -f logs/backend.log
# Follow veto decisions in real time
tail -f logs/veto.log
# Follow trade opens and closes
tail -f logs/trades.log
During initial deployment, watch logs/session.log at startup to verify that all 18 SMK modules loaded without import errors. Each module emits a [INFO] or [WARN] line as it initializes.
System status endpoint
GET /api/status returns a comprehensive snapshot of the pipeline state, including module counts, live feed status, and bar count:
curl http://localhost:8000/api/status
{
"status": "ready",
"bars_loaded": 2340,
"symbol": "EURUSD",
"pipeline": "initialized",
"modules_ok": 18,
"import_errors": 0,
"plugins": 6,
"live_feed": {
"running": false,
"symbol": null,
"granularity": null,
"clients": 0
},
"aegis": {
"slm": true,
"aegis": true,
"router": true
}
}