Aptos has introduced a set of AI development tools aimed at improving how coding assistants interact with the Move programming language. The tooling addresses a longstanding challenge: AI models trained primarily on JavaScript and Python have struggled to work effectively with Move, a niche language used on the Aptos blockchain.
What Aptos Released
The centerpiece is the Aptos Model Context Protocol (MCP) server, detailed in Aptos documentation and updated on October 9, 2026. The server enables AI tools to call Aptos APIs and query on-chain data using natural language instructions. Developers can describe their needs in plain English, and the assistant handles the underlying API work. The server also supports code generation and project scaffolding through a single npx command installation.
The MCP server is compatible with several widely used AI coding tools, including Claude Code, Cursor, OpenAI Codex, and GitHub Copilot.
The second component is a set of standardized LLMs.txt feeds that convert Aptos documentation into machine-readable format for large language models. The feeds are available in full and condensed versions.
The third piece is Aptos Agent Skills, updated as of September 14, 2026. According to documentation, the skills cover writing secure Move V2 contracts, running security audits, and generating tests with 100% coverage.
Why Move Required New Tools
Move was originally developed for the Diem project at Facebook. After that initiative concluded, the language found continued use on the Aptos blockchain.
Move is built around resource safety and formal verification. Resource safety treats digital assets like physical objects—they cannot be accidentally copied or deleted. Formal verification allows code to be mathematically checked to confirm it behaves as intended.
Aptos previously partnered with FLock.io in 2024 to refine how AI models interact with Move. The new tooling builds on that earlier work.
Implications for Developers
For developers, the practical benefit is a faster onboarding process. Rather than manually feeding documentation to a chatbot, builders can point an AI assistant at Aptos to query live chain data, scaffold projects, and draft contracts.
Given that smart contract bugs are permanent and costly once deployed on a blockchain, tooling that encourages AI-generated code to follow secure Move V2 patterns, include audits, and maintain full test coverage addresses known failure points in crypto projects.
However, AI-generated code depends on the quality of human review it receives. While agent skills emphasize audits and testing to mitigate risks, they do not replace direct human oversight of contract code.
