Each agent answers a different question about your own trading data. This is the detail behind the summary list: what each one needs before it can say anything useful, and what a run costs against your allowance.
Market Briefing
Pre-session scenarios, key levels and the economic events that matter that day. It does not need your trades.
It is the one weighted agent: a briefing is a multi-step debate rather than a single answer, so it is charged at 10 runs instead of 1. With no active plan the daily allowance is 1, which means the briefing is not reachable there at all.
Trade Coach
Grades a closed trade A to D with the coaching to go with it. It needs closed trades, and it runs automatically on newly closed ones, so most of the time you read it rather than start it.
Rule Enforcer
The end-of-session pass over your own rules: which ones you broke, on which trades, and your adherence figure. It needs rules written down and trades to judge, and it only judges days from the moment each rule started applying.
Risk Guardian
How close each account is to its daily loss limit and its drawdown, measured per account against that account's own limit. It needs a daily loss limit set, which is why the setup checklist asks for one.
Growth Coach
The weekly view: what the week did, what to keep, and what to change next week.
Psychology Tracker
The behavioural patterns your history shows, such as what your win rate does after a loss. You do not operate it from the agents list. Writing a reflection on a calendar day is the operation, and its thirty-day output renders on the month.
Coach chat
Free-form questions about your own record. One reply costs one run.
What every one of them needs
Trades. The agents read your data and nothing else, so with no imported or synced trades they say so instead of inventing something. Anything account-specific also needs the account connected, or the trades imported and attributed to it.
New agent runs and new coaching conversations need an active plan. Runs you already have stay readable whatever happens to the plan.
What a run costs
One run is one AI generation made on your behalf. Everything is charged at 1 except the Market Briefing, which is 10. Work the server starts on its own is not charged to you. The per-plan daily and monthly allowances, and when they reset, are covered in the article on AI usage limits.