AM

AI Master

Decision log

OverviewAssistantBoardInstrumentsPerformanceMarketDecisionsAI CallsSignalsWatchlistEventsHealthAgentsPromptsControl

View only. Nothing here can change the engine.

Idle

Agents

Pick a model from the list, or leave it on engine default. The scanner runs at its daily clock times, not when the engine starts. A price watcher can re-run it only after a level actually breaks. Risk limits stay in the gate.

Scannerscannergpt-6.1-sol

One reasoning call that compares every enabled instrument and puts 0 to 3 on the board. Code filters dead days first. A watcher re-runs it on a big 5-minute candle or when a level it set breaks - not on a clock.

0 calls 24h0 tokens

When it runs

Not when the engine starts. Daily at 07:00 and 13:00 broker time. No repeating timer.

Price watcher is on: a big 5-minute candle or a break of a level it set re-runs it, at most every 20 minutes, 6 AI calls per hour.

Last scan: watcher: level: NDX100 leaves 30953.6-31045.0 (Below the Asian/day low, bearish continuation gains room toward yesterday's low at 30897.4. Above the recent recovery high, the immediate downside setup weakens.), price 30933.8 · 2 weeks ago · board GBPUSD

Empty uses gpt-6.1-sol (SCANNER_MODEL). This call has no tools, so it can reason.

JSON repair and, for the analyst, the research brief. Empty uses AI_FAST_MODEL (gpt-6-luna).

The scanner does not run when the engine starts. It runs at the daily slots below.

Comma-separated HH:MM. Empty means no clock slots.

0 = off. Price, not a timer, should retrigger the scanner.

0–3 instruments the scan may pick.

Minimum time between watcher-triggered scans.

Price watcher

Re-run on a big M5 candle or a level break.

Prefetch news

One web search per eligible instrument before the call.

default applies to every instrument; add a symbol block for price units, e.g. "XAUUSD": {"min_day_range": 15, "m5_move": 10}.

Prompts
Analystanalyst

Looks at one board instrument while it is flat and decides whether to arm conditions or propose a trade.

0 calls 24h0 tokens

Empty uses AI_MODEL from .env (gpt-6-sol).

JSON repair and, for the analyst, the research brief. Empty uses AI_FAST_MODEL (gpt-6-luna).

research_ttl_minutes: how long a research brief is reused (it is also rewritten on a new day if research_new_day, after high-impact news if research_on_news, or when the model asks - no sooner than research_min_gap_minutes).

Prompts
Trade managertrade_manager

Runs while a trade is open: trails the stop toward the target, moves the target, takes partial profit or closes. A stop can never be moved toward a bigger loss.

0 calls 24h0 tokens

Empty uses AI_MODEL from .env (gpt-6-sol).

JSON repair and, for the analyst, the research brief. Empty uses AI_FAST_MODEL (gpt-6-luna).

Prompts
Owner assistantowner

Your assistant on the Assistant page. It reads the market, arms directives and hands trades to the engine. Every trade still passes the gate.

0 calls 24h0 tokens

Empty uses AI_MODEL from .env (gpt-6-sol).

JSON repair and, for the analyst, the research brief. Empty uses AI_FAST_MODEL (gpt-6-luna).

allow_trades: false makes the assistant read-only - it can still arm notify and ask_analyst directives.

Prompts
Reviewerreviewer

Grades closed trades and writes a daily review. Its lessons only reach the prompts once you approve them on Performance.

0 calls 24h0 tokens

Empty uses AI_MODEL from .env (gpt-6-sol).

JSON repair and, for the analyst, the research brief. Empty uses AI_FAST_MODEL (gpt-6-luna).

Comma-separated HH:MM. Empty means no clock slots.

How often closed trades are graded.

batch is how many closed trades are graded per run.

Prompts