About PlumBoard
PlumBoard is a task management system for humans and agents.
Design principles
Simple
Create, update, and delete tasks. Assign tasks to owners. Track task status. Add comments and attachments.
Ergonomic for humans
Web app with a visually simple graphic user interface that is easy to use and empowers advanced users with keyboard shortcuts.
Ergonomic for agents
Command line interface that supports structured JSON output, full UUIDs, schema introspection, dry runs, and field filtering, as well as an accompanying skill file that guides agents on how to use the CLI.
Open source
The code is open source and available on GitHub.
Why I built it
1. I want cloud-based state for local agent work
Too much planning and context in coding-agent workflows is session-scoped and locally scoped. That is inconvenient when the work should persist, be shareable, or remain available across machines. Extending PlumBoard into a cloud-based state layer for ongoing agent sessions solved a real problem I had myself.
2. CLIs + skill files feel like an important new interface
A GUI is still the right interface for humans (I think?), but a CLI plus a skill file may be the right interface for software's newest and soon-to-be biggest user groups: agents.
Building PlumBoard to work seamlessly for agents via a CLI with skill file instructions was both practically useful and a fun way to get hands-on experience with this new interface pattern for agent-driven software.
One article that helped shape that thinking, and CLI design, was "Rewrite your CLI for AI agents" by Justin Poehnelt.
3. The human-agent collaboration layer is interesting
What I find most interesting is designing a product at the intersection of human and agent collaboration. At its core, PlumBoard is a task management system that can work among humans, among agents, or across both at once. The GUI serves human workflows. The CLI serves agent workflows. That dual interface feels like a pattern we are likely to see much more of, and PlumBoard has been a useful way to explore it in practice.
