---
url: /mcp/tools.md
description: >-
  The 20 core TTMT MCP tools plus three default-enabled market-analysis tools,
  with read-only access and explicit evidence limits.
---

# What Your AI Assistant Can Read — MCP Tool Catalog

When you connect Claude, ChatGPT, or any other MCP client to TTMT, it gets a set of **tools** — small, named actions it can call to fetch a slice of your trading data and answer a question in plain language. This page lists all of them so you know exactly what your assistant can see when you ask it something.

::: warning Everything here is read-only
Every tool below only **reads** your data. Nothing on this page can place a trade, modify or close a position, change a setting, or move funds — and none of it can touch another user's data. There is no write tool, and no plans you make with your assistant take effect until *you* apply them yourself in the dashboard. If it ever says it "placed" or "closed" a trade for you, that is a mistake — it cannot.
:::

## The 20 core tools

Each tool answers a specific kind of question. You never call these by name — you just ask a normal question, and your assistant picks the right tool (or chains several) on its own. The example column shows the sort of thing that triggers each one.

| Tool | What it does | Example question |
|:---|:---|:---|
| `list_accounts` | Lists your connected MetaTrader accounts with id, label, currency, type, and current halt state. | "Which of my accounts are halted right now?" |
| `list_trades` | Lists trades with account, channel, status, symbol, and date filters. Paginated, newest first. The compact row does not carry the price-provenance record — `get_trade` does. | "Show me my losing `XAUUSD` trades this week." |
| `get_trade` | Pulls one trade by its `trd_*` reference or UUID — the signal, its positions and orders, a reconstructed lifecycle timeline, and where each executed price came from (see [Signal Sanitizer](/safety-and-reliability/signal-sanitizer#what-you-ll-see)). | "Walk me through trade `trd_abc123`." |
| `get_trade_trace` | The deep dive: a compressed, redacted, step-by-step engine trace for one trade or signal (`trd_*` / `sig_*`), for when the timeline alone doesn't explain *why* something happened. | "Why did `trd_abc123` only fill 2 of 4 layers?" |
| `list_positions` | Your currently open and partially closed positions, with a server-computed total unrealized P\&L and a by-symbol breakdown. | "What's my total floating P\&L right now?" |
| `get_performance` | Aggregated closed-trade performance — net P\&L, win rate, profit factor, average realized R — optionally grouped by account, channel, preset, symbol, or month. | "How does my Aggressive preset compare to Balanced last month?" |
| `get_settings` | Resolves your effective settings across both layers (your base row plus any channel-bound preset override), naming which layer supplied each field. On a channel with topic rules, the preset it uses is the channel preset. | "What settings apply when GoldFX signals hit my live account?" |
| `lookup_setting` | Finds one trading setting by name — its key, label, database column, or engine name — for clients that can call tools but can't read MCP resources directly. Returns the same detail as reading that setting's own resource: its default, what values it accepts, whether the engine actually enforces it yet, and where to change it. | "What does the Breakeven setting actually do?" |
| `get_profiles` | Preset summaries, or one preset fully resolved so you can see every value it actually uses. The API and MCP tool names still say profile; in the app these are called presets. | "What does my Aggressive preset actually override?" |
| `get_digest` | A one-call activity rollup for a period: closed P\&L, opened trades, open positions, halted accounts, skipped signals, and your best and worst channel. | "Give me my morning trading summary." |
| `list_channels` | Lists your tracked channels with their per-account routing assignments and bound presets. A channel on an account can have topic rules; the preset shown is the channel preset, and topic rules are not listed. | "Which channels route to my demo account?" |
| `list_signals` | Lists parsed signals — including the ones that were **not** traded (rejected, ignored, or skipped). Posts skipped by topic rules on every account do not appear here. | "Why didn't my last `EURUSD` signal get traded?" |
| `get_risk_status` | Per-account daily-loss-halt state plus the recent risk-limit audit trail. | "Show me the halt history on my live account this week." |
| `settings_healthcheck` | An advisory review that pairs your settings with your realized per-preset and per-channel performance and flags anything mechanically off. Never a backtest. | "Is my current config actually working well?" |
| `preview_execution` | Given a hypothetical signal and your current (or a named) settings, returns the exact adaptive order plan the platform *would* produce. Pure computation — **no trade is placed**. | "If I got a `XAUUSD` `BUY` signal now, what would my order plan look like?" |
| `get_market_status` | Whether the forex market is open right now and the next open/close time — DST-aware. Handy when nothing seems to be trading. | "Is the market open right now?" |
| `get_drawdown` | Realized equity-curve stats from your closed trades: maximum and current drawdown, longest win/loss streaks, and how long recovery took. | "What's my max drawdown, and how long did it take to recover?" |
| `get_calendar_performance` | Closed-trade performance bucketed by weekday, hour, or trading session (Sydney / Tokyo / London / New York), so you can see *when* you trade best. | "Am I a better trader in the London or the New York session?" |
| `get_exposure` | Your open positions broken down by currency — net long/short exposure per currency, per-symbol subtotals, and a warning when too much sits on one leg. | "How exposed am I to the US dollar right now?" |
| `explain_skip` | A one-call explanation of why one signal was or wasn't traded — its status, the channel's routing, any risk halts around that time, and the engine trace when available. An account that a topic rule skipped is named with its reason. | "Why exactly didn't my last `EURUSD` signal get traded?" |

::: info About `preview_execution`
`preview_execution` describes the order plan for a hypothetical signal using current or selected settings. It does not replay historical market prices or calculate a counterfactual return. Nothing is sent to your broker and no order is created.
:::

## Resources (7)

These are the base connector resources. The separately gated extension below adds an eighth.

Alongside the tools, the connector exposes seven **resources** — reference material your assistant can pull in to ground its answers in TTMT's own terms rather than guessing.

| Resource | What it holds |
|:---|:---|
| `ttmt://glossary` | Plain-language explanations of TTMT concepts — layered entries, take-profit redistribution, realized R, effective settings, halts, broadcast mode, soft delete. |
| `ttmt://settings/schema` | A field-by-field reference for your trading settings: the adaptive order engine, TP strategy, risk limits, breakeven, trailing stop, and the two-layer inheritance model. |
| `ttmt://settings/{dimension}/{key}` | One setting's own reference — its default, what values it accepts, whether the engine actually enforces it yet, where to change it in the dashboard, and links to any guide that covers it. |
| `ttmt://guides/{slug}` | One of TTMT's "Ideal Settings" guides — the same configuration walkthroughs published in these docs — so your assistant can ground a recommendation in the platform's own advice instead of improvising one. |
| `ttmt://accounts` | Your connected accounts with their current halt state — the same data as `list_accounts`, re-fetched fresh on every read. |
| `ttmt://trades/{ref}` | One trade by its `trd_*` reference or UUID — the same detail as `get_trade`, so your client can link straight to a single trade. |
| `ttmt://channels/{name}` | One tracked channel by name — its per-account routing assignments and bound preset (the channel preset; topic rules are not listed). |

These are read-only too. They give your assistant the vocabulary and structure it needs to explain your numbers correctly.

::: info Settings and guide reads count toward your daily limit
Reading a specific setting (`ttmt://settings/{dimension}/{key}`) or a guide (`ttmt://guides/{slug}`) counts toward the Essential plan's daily allowance, the same as a tool call — asking your assistant to look something up in the settings reference is itself one of your day's uses. The glossary, your accounts, a trade, and a channel stay free to read no matter how many times it reaches for them. See [Daily limits by plan](/mcp/security#daily-limits-by-plan) for the full picture.
:::

## Experimental market-analysis extension

::: warning Default enabled; operators can disable it
The extension adds three tools, one viewer resource and one prompt: 23 tools, 8 resources and 8 prompts in total. Provider access, historical evidence and client UI support can limit what you can use. A default-enabled tool is not proof of provider readiness or production rollout. See [Market history & what-if analysis](/mcp/market-analysis).
:::

| Tool | What it does | Example question |
|:---|:---|:---|
| `get_price_history` | Genuine external LSE candles around your trade/signal or a symbol and UTC window, with coverage and provenance. External prices are not broker-fill evidence. | “Show the market around this trade, separately from its recorded trace.” |
| `replay_history` | Paired historical baseline and explicit alternatives on the same eligible trades and frozen data, with exclusions, no-effect receipts and sensitivity. | “For Channel X last week on Demo, compare replacing TP1 with 150 pips. Show what the evidence cannot establish.” |
| `show_analysis` | Inspect an existing private result: summary, comparison, trade, candles, simulated events, recorded trace or bounded CSV. No new calculation or provider fetch. | “Show the comparison and explain this trade’s recorded versus simulated exits.” |

The extra resource is `ui://ttmt/market-analysis-v1.html`, an optional MCP App viewer. The extra prompt is `/ttmt-market-analysis`. Without an interactive viewer, the assistant can still give the same text summary and tables. New history/replay calls count toward your daily allowance; existing-result inspection does not.

## Want a guided starting point?

You don't have to know which tool to reach for. The connector also ships seven ready-made **prompts** — one-click starting points like a morning review, a risk review, or a channel deep-dive — that chain the right tools together for you. See [MCP Prompts](/mcp/prompts) for what each one does and how to run it.

## Related pages

* [MCP Prompts](/mcp/prompts) — the seven guided prompts and when to use each.
* [MCP Security & Access](/mcp/security) — how read-only access is enforced, sign-in, and rate limits.
