fenbs for developers: one board for you, your users and your AI
A board with the three things you actually track — features, enhancements, bugs — refs you can put in a commit, roles for the non-developers, and a seat for Claude Code or Cursor.
Last checked
What it is
A board with four lanes and three kinds of task. Every task is a Feature, an Enhancement or a Bug and gets a reference — FE-014, EN-015, BU-016 — that never changes and is never reused, so it can go in a commit message, a branch name or a message to a client and still mean the same thing in a year. There are no epics, sprints, story points or custom fields, and there will not be. If you have found that a tracker with all of those ends up as a to-do list with extra steps, this is the to-do list without the steps.
The AI assistant is a member
Claude Code, Cursor, Copilot in VS Code, Claude and ChatGPT connect over MCP with a browser sign-in — no API key to paste — and hold your role, narrowed by the scopes you tick. Ask Claude Code to run the tests and file each failure as a Bug; ask Cursor to move the task it just finished and comment with the pull request. Every change carries the assistant’s name in the history, so you can review a day of agent work without reading a transcript. Connect steps: Claude Code, Cursor, Copilot.
AI context lives on the board
The notes an assistant should read before it touches your code — how you name things, what must never be done, where things are — are kept on the board, per project, rather than inside one assistant’s memory. Switch assistant tomorrow and the new one reads all of it on arrival. Assistants can add to it, signed.
The people who are not developers
The client, the designer, the person who tests things: each gets a role that says what they can do. The smallest role lets someone report bugs and nothing else. Refusals are sentences — “You are a Viewer on Kitchen Refit, which cannot move tasks between lanes” — so nobody has to ask you why a button is missing.
What you give up
GitHub integration beyond what an assistant does for you; sprints, estimates and burndowns; custom workflows. If your team runs on those, Linear or Jira is the right call and this page will not argue.
How to try it
Do not evaluate a simple tool with a demo project. Paste in the real list you are working from this week using Add many, add the one person who most often tells you about problems with a small role, connect the AI assistant you already use, and work from the board for a week. At the end ask one question: is the board still true? Your own board is free, so the trial costs nothing but the week.