> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.ninjatrader.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.ninjatrader.com/_mcp/server.

# Workflows

**Beta, pre-release.** The NinjaTrader MCP server and its trading skills are in beta and
aren't yet generally available. The beta runs on Demo only, which trades simulated money.
Live access opens later. Server hostnames, the tool set, the skills, and the onboarding
flow may change before launch.

Every workflow below follows the same contract: you ask in plain English, the skills gather data and do the math, and anything that would change your account arrives as a proposal for you to approve. See the [Skill Reference](/mcp/skills-reference) for what each skill does on its own.

## Size and bracket a new trade

> "Buy ES."
>
> "Size a trade on NQ with a 30-point stop."
>
> "How many contracts?"

**What happens:** `contract-intel` resolves a bare product code like "ES" to the actual front-month contract. `pretrade-risk` then sizes the trade from your risk budget and stop distance—pulling ATR from `market-context` when the stop is volatility-based—and proposes an OCO bracket: a stop plus a profit target at a multiple of your stop distance. You get the contract count, the dollar risk, and a daily loss-budget check, and `estimate_order` runs as the final feasibility gate. A setup-quality question such as "Is this a good setup?" routes to `risk-coach` instead. That skill answers with behavioral context from your own history, not a sized bracket.

**What you approve:** the order payload. The skills never submit it—you do.

## Monitor a live position

> "Position health."
>
> "Margin usage."
>
> "Daily loss."

**What happens:** `position-watchdog` reads your open positions and working orders, then reports open P\&L in dollars and R, distance to your stop in ticks and dollars, progress toward your target, margin usage, and guardrail flags, including whether you've moved your stop since entry. It pulls `market-context` for regime context when narrating position health, and it can run what-if math on a stop move without touching anything.

**What you approve:** nothing. This workflow only reads data. To actually move the stop, ask, and `scale-manager` proposes the payload.

## Scale in or out

> "Take half off."
>
> "Move stop to breakeven."
>
> "Ladder into ES at 7148 / 7145 / 7142."

**What happens:** `scale-manager` computes your weighted cost basis after the change, your dollar risk before and after, and the exact order payloads—modify, place, or close—for the move you described. Ladder entries support front-, back-, or equal-weighted splits.

**What you approve:** every payload, one by one.

## Set an alert

> "Alert."
>
> "Notify."
>
> "Set a trigger."

**What happens:** `alerts-composer` translates your intent into a validated alert expression—combining price, position P\&L, account equity, margin, and session levels—checks the syntax offline, restates the alert in plain English, and checks for duplicates before submitting through `create_alert`. `pretrade-risk`, `event-watch`, and `position-watchdog` propose alert expressions through the same path, such as a reaction-window bracket around a CPI release. Once submitted, alerts fire inside the NinjaTrader and Tradovate platforms—in-app delivery is always on, and the mobile apps can push them to your phone.

**What you approve:** the restated alert before it's submitted.

## Review your day

> "Best and worst trades?"
>
> "What went wrong?"
>
> "What if I'd used a trailing stop?"

**What happens:** three skills split the question. `trade-journal` pairs your fills into round-trip trades and reports P\&L, win rate, and hold times by symbol, P\&L and trade count by hour, and per-fill slippage—the descriptive "what happened." `trade-debrief` consumes that round-trip pairing and mines the session for mistakes, patterns, and correction rules, citing every claim to the underlying report and leading with dollar impact—proposed changes are hypotheses, not personalized advice. And `trade-replay` deep-dives one trade: how far it ran for and against you, and what alternate exits—a bigger target, a trailing stop, a breakeven pin—would have done against the same bars.

**What you approve:** nothing. The review workflows only read data.

## Chart It

> "Show the chart."
>
> "Plot."
>
> "Render."

**What happens:** `chart-render` draws three kinds of static PNG charts. The first is candlesticks with VWAP, fill markers, and price-level overlays. The second is a volume-profile histogram with POC and value-area lines. The third is an equity curve with drawdown shading. Bars come from `market_history` and fills from `fill_history`, with analytics from `market-context` and round-trip trades from `trade-journal` when a chart needs them. Requires `matplotlib`.

**What you approve:** nothing. Charting only reads data.