Axiomkit | AI Agent Typescript Framework

Project Description

AxiomKit is a TypeScript-first framework for composing AI agents and connecting them to real-world systems. It’s designed for clarity, reuse, and speed-to-demo: a clean core, batteries-included integrations, and runnable examples you can adapt in minutes. The project ships as a monorepo with a core engine, a CLI, and adapters for popular platforms like Discord, Telegram, Supabase, MongoDB, Sei, and MCP- plus a polished docs site built with Next.js.

What the Project Does

• Enables developers to define agents with structured prompts, tools, memory, and execution flows.
• Provides an opinionated core for tasks, handlers, and services with clear interfaces.
• Offers ready-to-run examples for chat bots, memory patterns, blockchain queries (Sei), and more.
• Includes a CLI for scaffolding agents and managing dev ergonomics.
• Publishes a documentation app with guides, examples, and live code snippets.

My Responsibilities

Describe the challenges faced during this project.

Key Technologies Utilized

• Frontend: Next.js 15 , Fuma docs ,MDX (content)
• Backend: Node.js, pnpm (monorepo) , vitest.
• Database: MongoDB , Supabase , In-memory storage
• Intergrations & APIs: Discord.js , Telegram Bot API , MCP ,Supabase client.

Conclusion

AxiomKit is my answer to building AI agents that are both production-minded and easy to learn. It balances strong core primitives with pragmatic integrations, so teams can prototype quickly without painting themselves into a corner.