Hi everyone, and please check out my project Quarterdeck
Hi all, I'm Shannon. Lately I've been building tools that give AI agents work that keeps going, not just one-off tasks.
I'm a developer who loves working with all things AI, whether it be fine-tuning a model, wrangling agents or hooking an app up to an API.
My current project is Quarterdeck, a Claude Code plugin for persistent goals. These are standing jobs that never finish, like "keep my inbox under control" or "pick up new GitHub issues." The agent doesn't hold the goal's state. Quarterdeck keeps it in local SQLite: instructions, open items, memory, your past decisions and run history. Each run starts in a background subagent, reads one brief, does the work and records a summary. You approve the actions it proposes from a Goals pane beside the chat. A guard plugin keeps each run inside its goal. Runs can start on a schedule or when something happens, like a new issue, a feed entry or a changed page.
I built it because every session started from scratch. I wanted agents to pick up where they left off and to ask before doing anything I hadn't approved.
Demo (a few minutes): youtube.com/watch?v=okKu9MnWZ5Y…