> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.6mm.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.6mm.com/_mcp/server.

# AI Hub Overview

6MM Agentic AI Hub is the AI-native interface for 6MM market data, account workflows, and future controlled trading operations.

Instead of switching between dashboards, API docs, and scripts, users can ask an AI client to query 6MM, explain market conditions, review account risk, or prepare an order preview. The AI client does not replace the 6MM API security model; it calls structured 6MM tools through CLI, MCP, or workflow templates.

<h2 id="what-you-can-ask-an-ai-client">
  What you can ask an AI client
</h2>

Use AI Hub when you want to ask questions such as:

```text
Show me the BTC-USDT-PERP order book and explain current liquidity.
```

```text
Compare BTC and ETH funding rates over the last 24 hours.
```

```text
Check my testnet positions and explain liquidation risk. Do not place orders.
```

```text
Preview a small BTC long order and explain fees, margin impact, and risk checks.
```

<h2 id="why-ai-agents-need-a-dedicated-access-layer">
  Why AI Agents need a dedicated access layer
</h2>

AI Agents work best when they have structured tools instead of raw webpages or unstructured copied text. AI Hub gives them:

| Requirement         | 6MM approach                                                                                       |
| ------------------- | -------------------------------------------------------------------------------------------------- |
| Live market context | Public market-data tools and resources for symbols, order books, trades, funding, fees, and rules. |
| Clear permissions   | Separate public data, account-read, preview, and controlled trading scopes.                        |
| Safer execution     | Preview-first order flow, explicit confirmation, idempotency, limits, and audit logs.              |
| Client flexibility  | Hosted MCP, Local MCP, CLI, Remote MCP, and Agentic Skills for different AI environments.          |

<h2 id="what-ai-hub-includes">
  What AI Hub includes
</h2>

Command-line access for terminals, scripts, and coding agents that can execute shell commands.

Run a local MCP server so AI clients can call 6MM tools while user credentials stay on the local machine.

Connect to a 6MM-hosted MCP endpoint for zero-install public market data and rule context.

Reusable workflow templates for market analysis, portfolio review, liquidation protection, and execution planning.

<h2 id="setup-paths">
  Setup paths
</h2>

| Path           | Best for                                            | User should know                                                                       |
| -------------- | --------------------------------------------------- | -------------------------------------------------------------------------------------- |
| Hosted MCP     | Quick public market-data access                     | No install and no user API key for public data. Trading is not the first-release goal. |
| Local MCP      | AI clients that need private account context        | The MCP server runs locally and credentials remain on the user's machine.              |
| CLI            | Coding agents, scripts, and terminal users          | Produces agent-friendly command output and can be used without an MCP server.          |
| Remote MCP     | Institutions and teams that want their own endpoint | The user or institution controls hosting, credentials, network access, and logs.       |
| Agentic Skills | Repeatable workflows                                | Skills guide the AI through analysis and risk steps using approved tools.              |

<h2 id="what-is-available-first">
  What is available first
</h2>

AI Hub should start with public market data and trading-rule resources:

* Markets.
* Tickers.
* Order books.
* Trades.
* Klines.
* Funding rates.
* System status.
* Fees, contract specs, margin rules, and risk rule resources.

Private account access and trading should be staged:

* Account-read tools require explicit read-only credentials.
* Order preview should come before any real order creation.
* Controlled trading requires permission scopes, service-side risk checks, user confirmation, idempotency, and audit logs.

<h2 id="start-here">
  Start here
</h2>

Pick the right path based on your AI client, security needs, and deployment model.

Start with ChatGPT Desktop, Claude Desktop, Claude Code, Codex CLI, Cursor, Gemini CLI, VS Code, or Windsurf.

Start with public market data without installing a local server.

Use a local server when private account context or future trading tools are needed.