2026 / active
Lolo IRC Bot
A modular IRC system that brings AI conversations and useful tools into a long-running real-time chat environment.
Lolo explores what a modern, tool-using assistant looks like inside IRC. It supports conversational AI, web and URL analysis, image operations, code execution, and persistent user context without losing the conventions of an established chat protocol.
The system
A Go IRC client handles connection state and chat events while a separate AI service owns model interaction and external tools. Permission levels, moderation controls, rate limits, and audit logging keep powerful capabilities bounded in shared channels.
What I focused on
The engineering challenge is reliability at the seams: reconnecting cleanly, preserving conversational context, separating tools from transport, and ensuring an AI response cannot bypass channel permissions. The project has evolved through several architectures as I have tested those boundaries in real use.