Manage a product backlog with AI
Keep requests, priorities and acceptance criteria together in a Markdown backlog, with a separate record of completed or cancelled work.
Manage a product backlog with AI when requests arrive faster than the team can discuss them. The manage-backlog skill turns agreed scope into an editable Markdown file or updates the one you already use. Each item has an identity that survives a change of priority.
Suppose a customer asks for “better receipts”. Before anyone books development time, decide whether that means a printable copy, clearer tax details or both. The backlog gives each agreed change a test and keeps unresolved questions visible until somebody can answer them.
What it does
- Record requests as features, fixes, investigations or tasks.
- Write acceptance criteria that a second person can independently check.
- Distinguish a launch requirement from a worthwhile improvement that could wait.
- Break oversized requests into pieces that each deliver a useful result.
- Keep an archived item’s ID so later discussions can refer back to the decision.
Install and use
- Install the skill in a file-capable agent and provide the agreed scope or your existing backlog. No tracker account is required.
- Ask: “Sort these requests for the next release. Leave anything with an unresolved dependency out of the ready list.”
- Check estimates with the team and review any proposed splits before saving the revised backlog.
FAQ
Source
Made by Philipp Thoss. Original skill: manage-backlog, published on GitHub under the MIT licence.
View on GitHub