AI Agent Integration
How AI agents can issue and manage regulated tokens via Sinai Standard
AI Agent Integration
AI agents are increasingly used to automate financial operations — from treasury management to programmatic token issuance. But autonomous agents operating in regulated environments need guardrails built into the infrastructure itself, not bolted on as afterthoughts.
Sinai Standard provides that infrastructure. Every token issued through the protocol carries embedded compliance rules (KYC allowlists, transfer taxes, hold periods, balance caps) that are enforced on-chain. This means an AI agent cannot accidentally mint tokens to a non-KYC'd wallet or bypass a hold period — the transaction simply fails at the protocol level.
Integration Paths
There are three ways to connect an AI agent to Sinai Standard:
| Approach | Best For | Setup | Flexibility |
|---|---|---|---|
| MCP Server | Claude, GPT, and other LLM-based agents | Minimal — config file only | Tools are pre-defined |
| Agent SDK | Custom autonomous agents | Low — one npm install | Full method access, JSON-friendly |
| Raw AksumKit | Advanced integrations | Medium — direct SDK usage | Maximum control |
MCP Server
The Model Context Protocol server exposes Sinai Standard as a set of tools that any MCP-compatible AI agent can call. No code required — just configure the server and the agent can issue tokens, manage allowlists, and query compliance state through natural language.
Best when you want an LLM to interact with Sinai Standard directly.
Agent SDK
The Agent SDK wraps AksumKit in a JSON-friendly interface designed for programmatic agent consumption. All methods return structured { success, data, error } responses and never throw exceptions, making it safe for autonomous operation.
Best when you are building a custom agent that needs deterministic, programmatic access.
Raw AksumKit
For maximum control, agents can use AksumKit directly. This is the same SDK used by human developers and provides full access to every protocol feature. The tradeoff is that error handling, serialization, and transaction management are your responsibility.
Best when your agent framework already handles Solana transaction lifecycle.
When to Use Which
- You have a Claude or GPT agent and want it to manage tokens → MCP Server
- You are building an autonomous agent in TypeScript/JavaScript → Agent SDK
- You need fine-grained control over transactions and accounts → Raw AksumKit
Note: All three paths enforce the same on-chain compliance rules. The choice is about developer experience, not security posture.