System

System configuration, agent topology, and service health.

Doctrine self-healing

Optional automation that closes positions on your behalf to clear rule breaches. Default OFF — every position close is manual unless you arm this.

Auto-Reconcile to Doctrine

When ON, the platform closes positions on your behalf at the next 09:30–15:30 ET scan to clear every doctrine breach (over count, heat above 2.2%, missing brackets, RSI-2 shorts, off-universe symbols). You see statistics + the cycle summary; no per-position clicks. Default OFF preserves your full manual control.

DISARMED

Multi-Agent Decision Flow

Data flows from sources → orchestrator → specialist agents → trade decisions → execution → reflection loop.

Model Configuration (— Agents)

Click any agent to see its system prompt and available tools.

AgentModelTierTools
sentinel_hub (orchestrator) FAST
market_analyst LITE
macro_sentinel LITE
strategy_advisor FAST
risk_manager LITE
trade_executor LITE
reflection_agent FAST
global_macro FAST
elastic_analyst LITE

Scheduler Events

Hourly Market Scan

10:30-15:30 ET (6x/day)

GET /market/scan?sensitivity=normal

Macro Dislocation Check

Every 15 min (market hrs)

GET /market/sectors/dislocations

EOD Reflection Digest

16:30 ET weekdays

POST /reflection/digest

Weekly Universe Screen

Sunday 18:00 ET

GET /market/scan?sensitivity=wide

Strategy universe

Override the platform's compiled-in 100-ETF default. Validated against Alpaca's live tradable-asset cache. Picks up the override on the next scan; no restart needed.

Strategy universe — operator override

Daemon reads this on every scan. When active, your list wins over the compiled-in 100-ETF default. Tickers validated against the live Alpaca asset cache (APP-025).

Loading override state…

Active Integrations

MongoDB AtlasOFFLINE

Persistent Memory — trade decisions, outcomes, snapshots, agent reflections

14 FastAPI endpoints (store, query, aggregate, sync)

Arize PhoenixACTIVE

Observability — traces, drift detection, LLM-as-judge

OpenTelemetry traces for agent self-improvement loop

https://app.phoenix.arize.com/s/mikey2k

OTLP HTTP collector + REST query API (Arize Cloud)

Alpaca MarketsACTIVE

Trading Execution — real orders on paper account

Paper trading — bracket orders, positions, market data

https://paper-api.alpaca.markets/v2

7 FastAPI endpoints (account, positions, orders, risk, news)

Google GeminiACTIVE

LLM Inference — agent reasoning and decision-making

— agents — Flash (orchestration) + Flash Lite (analysis)

https://generativelanguage.googleapis.com/v1beta

~42 LLM calls/day across — agents

Strategy parameters are now managed per-strategy on the My Strategies page. Each strategy has its own indicators, entry/exit rules, and risk parameters.

AI Strategy Advisor

agents · tools · Gemini Flash

Welcome to SentinelHub. I'm the AI Strategy Advisor — I coordinate 9 specialist agents to help you analyse markets, build strategies, and manage risk.

Ask me about market conditions, your portfolio, trading strategies, or past performance.

Powered by Gemini Flash · Connects to Alpaca, MongoDB, Phoenix