Start from the Bug Tracker project template in Manager: collect bugs and feature requests, grade them by severity, and let your agents reproduce and fix what they can.
Bugs and feature requests have their own types and workflows, so a crash never drowns in nice-to-haves.
Each bug can record how bad it is and where it happened, next to how soon it should be fixed.
Hand a confirmed bug to a coding agent over MCP. It reproduces the problem, fixes it and reports back on the bug.
Every report goes through triage: is it real, is it new, how bad is it? Severity measures impact, priority decides the order of work. A minor bug on the signup page can outrank a major one nobody sees.
A bug gets fixed once someone can reproduce it, ideally from exact steps and the environment. Someone other than the author then verifies the fix. A bug that comes back is a regression.
Each step of the loop becomes a status, a field, a label or a link.
| Bug tracking | In Manager |
|---|---|
| Triage | New to Confirmed for real bugs; Closed or Won't Fix for the rest. |
| Severity | The Severity field, Blocker to Trivial. Add it as a list column to sort by it. |
| Priority | Manager's own priority, Critical to Low. The High Priority filter shows Critical and High. |
| Environment | The Environment text field: version, browser, device, staging or production. |
| Duplicate | A Duplicates link to the original, then close the copy. |
| Verification | Fixed to Closed after someone checks the fix where the bug was reported. |
| Regression | The Regression label, plus a Causes link from the item whose change broke it. |
Connect your AI agent over MCP and let it create the project from the Bug Tracker template.
Say to your agentManager: Set up a project for my product from the Bug Tracker template and ask me for its name and key.
Log bugs from testers, customers or your own agents, with steps to reproduce, severity and environment. Attach screenshots and logs.
Say to your agentManager: File the bug I describe and ask me for steps to reproduce, severity and environment if anything is missing.
Confirm what is real, link duplicates, close what is not a bug and set the priority, so the worst problems rise to the top.
Say to your agentManager: Triage the bugs in New: find duplicates, confirm the real ones and set priorities.
Assign bugs to your team or a coding agent; tasks under a bug split a bigger fix. When the agent finishes, the bug moves to Fixed.
Say to your agentManager: Take the most urgent confirmed bug, reproduce it first, then fix it and describe the fix on the bug.
Someone checks each fixed bug in the environment where it happened before it moves to Closed. Regressions get their label, so patterns show up.
Say to your agentManager: Which bugs are in Fixed and still need to be verified?
A project created from the Bug Tracker template starts with these work item types and statuses. You can adapt all of them to your team later.
A defect or unexpected behavior that needs to be fixed.
A suggestion for new functionality or an improvement.
A unit of work under a bug or feature request.
Run the tracker as its own project beside development. Testers and customers join it without seeing your backlog, and a bug can link to its fix in the other project.
Testers file bugs directly or through their own agent. A feedback form or crash reporter can file them over the HTTP API with a project-confined API key.
Fixed counts as done, so Open Bugs drops a bug once it is fixed. Filter by the status Fixed to find what still needs verifying.
Open Bugs lists every bug not yet fixed, severity and priority show what comes first, and each fix is written up on its bug.
Get early access