ai-tools

This MCP server bundle provides a standard input/output interface for AI apps to interact with Sui Network through a collection of pluggable tools. The built-in tools originate from @suiware/ai-tools and include capabilities such as suiAddressTool, suiWalletBalanceTool, suiTransferTool, suiSwapTool (mainnet only), suiStakeTool, and suiUnstakeTool. You can obtain all available tools using the provided getSuiwareAiTools() helper in your integration, which returns the full suite of tools you can enable or disable when configuring an AI agent. The MCP server exposes a structured API that lets your AI assistant query wallet balances, fetch addresses, perform transfers, stake/unstake assets, and interact with Sui-based operations via the configured tools. To use the tools, load the MCP server in your app and invoke the tool functions as defined by the MCP protocol, then have the agent interpret and present results back to the user.

Typical workflows include: initializing a session, listing available tools, choosing a tool to perform an action, providing the required parameters (e.g., destination address, amount, token type), executing the tool, and handling responses or errors. For example, you can query a wallet balance, then initiate a transfer or a swap, depending on the tool enabled for the current assistant context. The examples in the repository demonstrate both simple synchronous interactions (e.g., balance lookup) and interactive agents that stream or generate text while coordinating with multiple tools.

If you’re building an AI assistant, integrate the MCP server into your app as the backend tool registry. Expose tool discovery (getSuiwareAiTools()), and route tool invocations through the MCP interface so the AI model can request actions and receive structured results suitable for presentation to end users.

Prerequisites

• Node.js (v18+ recommended) and npm or pnpm
• Access to the repository containing the MCP server (suiware-ai-tools monorepo)

Installation steps

1. Clone the repository or download the package

• git clone https://github.com/your-org/suiware-ai-tools.git
• cd suiware-ai-tools

2. Install dependencies for the MCP server package

• npm install or
• pnpm install

3. Build or prepare the MCP server (if applicable)

• npm run build (If a build script is not defined for the MCP package, you can skip to running the server directly.)

4. Run the MCP server

• From the repository root (or navigate to the MCP package):
  • npm run start
  • or node packages/mcp/src/server.js

5. Connect your client or AI app to the MCP server endpoint as documented in the MCP package docs.

Notes

• Ensure environment variables (like API keys or network settings) are configured as needed by your tools. See additional_notes for common env vars and options.

Tips and common considerations:

• If your tools require network access or API keys (e.g., blockchain RPC endpoints, wallet providers), set them via environment variables or a config file recognized by the MCP server.
• The suiSwapTool is marked as mainnet-only; ensure you enable or disable it based on the target network to avoid unexpected errors.
• Use getSuiwareAiTools() to discover which tools are currently available and how to enable them in your AI agent.
• When deploying, consider rate limits and per-wallet permissions for sensitive actions like transfers or staking. Implement proper validation and user confirmation flows in your app.
• If you upgrade packages, re-run installation steps to ensure compatibility between the MCP server and the ai-tools package.
• For troubleshooting, check the MCP server logs for tool invocation errors, missing parameters, or network-related failures, and verify that all required environment variables are set correctly.