vercel-labs/personal-agent-template

What it solves

This project provides a ready-to-deploy template for creating a personal AI agent that maintains a consistent identity and memory across multiple communication channels. It eliminates the need to build the same agent logic from scratch for every platform, allowing users to fork and customize their own assistant.

How it works

The system uses the Eve agent framework to manage channels, tools, and skills. It is built with Nuxt for the UI and API, using Better Auth for user authentication and SQLite for data storage. The agent's long-term memory is injected into every session, ensuring context is preserved across web chat, Slack, and iMessage. It integrates with external tools like GitHub and Linear via Vercel Connect, enabling the agent to query real-time data from those services.

Who it’s for

Developers and individuals who want a personalized AI assistant that can access their professional tools (GitHub, Linear) and be accessible via their preferred messaging apps.

Highlights

• Multi-channel access: Chat with the same agent via a web interface, Slack DMs/mentions, and iMessage.
• User-approved memory: The agent can propose new facts to save to its long-term memory using a save_memory tool, but requires explicit user approval.
• Professional tool integration: Built-in support for querying repositories, pull requests, and issues in GitHub and Linear.
• Durable sessions: Threads persist across sessions, allowing for continuous conversations.
• Memory import: Ability to import existing memory from other assistants like ChatGPT or Claude.