Developer Overview
Orien has two developer extension systems: MCP and Skill. They solve different product problems and should not be mixed into one package model.
How to Choose
| Build | Use When | Product Surface |
|---|---|---|
| MCP Server | You want AI Chat to call external tools, data providers, news sources, research services, or operator utilities. | AI Chat and MCP Market |
| Skill | You want a trading Agent to run strategy logic through observe, plan, risk, and execute phases. | Trading Agents and Skills Market |
MCP in Orien
MCP is the extension layer for AI Chat. It is best for interactive workflows where the user asks questions, compares information, or requests tool-assisted analysis.
Good MCP examples in our business:
- CEX market data tools, such as OKX read-only market data
- news and research tools for a token or sector
- on-chain analytics and wallet context
- project research helpers
- risk dashboards or operator lookup tools
For trading-related MCP servers, prefer read-only tools by default. Write actions such as order placement, signing, or account mutation should require explicit confirmation and platform risk controls.
Skill in Orien
Skill is the strategy package for trading Agents. It belongs to repeatable execution workflows, not ad hoc chat tooling.
Good Skill examples in our business:
- Hyperliquid BTC perp strategy logic
- factor combinations for trend, funding, volume, and order book signals
- backtestable trading rules
- risk-gated order execution
- strategy packages that can run in paper, backtest, or live modes
Skills should be inspectable and code-first. LLMs can help scaffold strategy code, but the runtime contract should remain explicit.
Relationship to AI Chat
AI Chat can help operators interact with the platform, inspect Agent behavior, and call enabled MCP tools. It should stay as the natural-language interface over the trading system.
When the workflow becomes autonomous, repeatable, or execution-sensitive, move it into a Skill-backed Agent instead of keeping it as a chat-only tool.
Next Steps
- MCP Development — Build tools for AI Chat
- AI Chat Integration — Understand how MCP tools enter chat routing
- Skill Development — Build strategy packages for trading Agents
- Skill Runtime Reference — Define the Agent runtime contract