Skip to main content
AI coding assistants were not trained on the specifics of Sei. Without context, they give you generic Ethereum answers. These answers get gas patterns wrong, miss dual-address handling, and use outdated package names. They also make assumptions about finality that do not hold on a 400ms chain. Sei has two complementary tools to fix this.

​
Two tools, two roles

sei-skill

Knowledge injection. Teaches your AI assistant everything about Sei: architecture, precompiles, wallet patterns, gas behavior, and the ecosystem. As a result, every answer it gives is Sei-specific from the start.

MCP Server

Live blockchain access. Gives your AI assistant tools to read and write on-chain: query balances, send transactions, interact with contracts, and monitor network state in real time.
Use them together: sei-skill gives your assistant knowledge of Sei, and the MCP Server lets it act on Sei. For a single area of Sei, such as contracts or bridges, install a focused skill from the skills registry.

​
Build AI agents on Sei

Cambrian Agent Kit

LangChain-based SDK for building autonomous AI agents with DeFi integrations for staking, lending, swaps, and liquidity.

Agentic wallets

Wallet infrastructure designed for AI agents: programmable signing, session keys, and policy controls.

x402 payments

HTTP-native micropayments that let agents pay per inference for LLM APIs, data feeds, and other monetized services.