Getting Started

Install, configure a demo account, place a first trade, then move on to the AI assistant and intelligence.

English (current) · 简体中文

From installation to your first trade, then the AI assistant and systematic research — this guide walks through the core workflows of Desic Terminal.

1. What is Desic Terminal#

Desic Terminal is an AI-native trading terminal for OKX USDT perpetual contracts that unifies six workspaces in one real-time context:

Workspace	What you can do
Trading	Market, order book, depth, candles, order entry, position management
Pro Charts	Multi-timeframe multi-pane charts, 20+ built-in indicators, layers, drawings and quick trading
AI Assistant	Read market/account/intelligence evidence in conversation, produce opportunities and chart actions
AI Automation	Background AI Profiles that wake on conditions, with audit and position reviews
Systematic Strategy	Python strategy backtesting, parameter tuning and live Profiles (no Python install)
Market Intelligence	Time-series evidence: news, sentiment, flows, Smart Money and system stress

2. Installation#

Download the build for your platform from GitHub Releases:

Platform	Package
Windows x64	`Desic-Terminal_Windows-x64-setup.exe`
macOS Apple Silicon (M1+)	`Desic-Terminal_macOS-arm64.dmg`
macOS Intel	`Desic-Terminal_macOS-x64.dmg`

Requirements

Windows 10+ or macOS 12+
Network access to OKX and your chosen AI provider
No Python install needed — the systematic research runtime ships inside the installer

3. First Launch#

The main window opens in the trading workspace: title bar and watchlist on top, candles in the middle, order book and ticket on the right.
The left rail switches workspaces: Trading, Opportunities, AI Automation, Systematic Research, Intelligence, Data.
The notification center in the top-right collects in-app notifications; you can forward them to a Feishu bot in Settings.

4. Configure an OKX Account#

Settings → Account → Add account:

Create an API key on OKX with trading-related permissions only.
Fill in Key / Secret / Passphrase and choose the environment (demo or live).
After saving, the desktop app opens a dedicated Private WebSocket and syncs historical fills and orders.

Recommendation	Why
Start with demo	Validate contract sizes, margin modes, stops and notifications first
No withdrawal permission	API keys must not include withdrawal rights
Separate credentials	Demo and live each get their own keys

5. Get to Know the Trading Workspace#

The trading workspace: watchlist, candles, order book depth, trades and the ticket all share one symbol and account context.

Every panel in the trading workspace shares one symbol and account context:

Watchlist: switch symbols on the left, drag to reorder.
Order book / Depth: live levels and depth chart on the right.
Recent trades: latest trade tape.
Ticket: limit, market, algo orders; attach TP/SL or OCO.
Positions & orders: manage, amend, cancel and close in the tabs below.

All quantities are submitted in OKX contracts (lots), with coin amount, notional value, estimated margin, fees and stop risk shown alongside — so contracts, coins and USDT exposure are never confused.

Chart essentials

Action	How
Change timeframe	Chart toolbar (1m – 1M)
Add an indicator	Indicator button → pick and configure; multiple instances supported
Toggle layers	Layers button → indicators / price lines / drawings / analysis / fills / measure
Detached windows	Chart windows button → 1/2/3/4-pane layouts, each pane with its own symbol and timeframe
Quick trading	Right-click the chart or drag price lines to form trade intents directly

6. Place Your First Trade#

Example: a demo limit order.

Pick the demo account in the top account switcher.
In the ticket choose Limit, a direction (long/short), price and size; attach TP/SL if needed.
The preview dialog shows contract count, estimated margin and fees — verify, then submit.
The order appears under "Orders"; once filled it moves to "Positions" and the chart shows the position line, entry price line and TP/SL lines (all draggable).
Close from the positions row, or drag the chart position line to a target price for a quick close.

7. Configure the AI Assistant#

Settings → AI to configure a model service:

Pick a provider: OpenAI, Anthropic, Gemini, Grok, DeepSeek, Qwen, KIMI, Doubao, MiniMax, Zhipu, or any compatible API.
Fill in the API key and model name, save and run a connectivity test.
Optionally configure a locally installed official Codex CLI or Claude Code channel — requests still obey the terminal's tool permissions.

8. Analyze the Market with AI#

Open the AI trading assistant on the right:

Ask, for example:

text

   Read the live price, order book, 1-hour candles, funding rate and open interest
   of BTC-USDT-SWAP and give a concise analysis with data timestamps.

The assistant reads live market, account, orders, history, news and Smart Money evidence through allowlisted tools, streaming reasoning, tool calls and timing.
Tool results carry data time, source and freshness, so the model distinguishes facts, inference and data gaps.

Permission modes (chosen per session/configuration):

Mode	Read market & account	Trade opportunities	External trade side effects
`advisor`	Yes	No	Forbidden
`copilot`	Yes	Create/edit/reuse	User approval required
`limited_auto`	Yes	Frozen-candidate submission	Profile-authorized scope only

9. Custom Chart Indicators#

The indicator center ships MA, EMA, VWAP, Bollinger, Supertrend, Ichimoku, MACD, RSI, KDJ, ATR, ADX, Stochastic, CCI, ROC, Williams %R, MFI, OBV and more.

Custom indicators use a safe JSON DSL (no JavaScript execution, no file or network access):

Indicator center → New custom indicator.
Write expressions in the CodeMirror editor, or use the indicator AI assistant to discuss ideas — DSL is only generated when you explicitly ask to create or update.
After saving, the indicator passes schema validation and versioning and enters the library for any chart.

10. Market Intelligence#

Market intelligence: news and events, coin sentiment, open interest and taker flow, stored locally as attributable evidence.

The intelligence panel is a set of time-series evidence shared by manual workflows and AI tools:

News & events: list, full text, clustering, importance and multi-window market reactions
Sentiment: coin sentiment snapshots, trends, heat and long/short rankings
Economic calendar: weekly/monthly views, region, importance, previous/forecast/actual
Flow structure: price + OI combinations, taker flows, retail vs elite positioning
Smart Money: trader performance, positions, historical orders, aggregate signals and divergence
Derivatives state: funding, basis, liquidation samples, insurance fund, price limits and ADL

Every view discloses data time and coverage; missing or stale data is flagged explicitly — no old snapshots disguised as current state.

11. Next Steps#

Want AI to watch the market and review trades in the background → AI Automation Guide
Want to backtest, tune and run Python strategies → Systematic Strategy Guide
Want implementation details and safety boundaries → PRODUCT.md and the strategy protocol

View source on GitHub