Oct 28, 2026 | Written by Jon Leslie

TL;DR: Use your agents to clean up duplicate tickets, actions from feedback, nest bugs to their features, and scan where work is stuck.

If you've run production in Jira or most other tools, you know the drill. A new feature becomes a pile of tickets (e.g. in games an art asset needs concept art, modeling, animation, and gameplay, so it becomes four tickets, a few links, and eventually a clone or two). Each copy collects its own notes and its own comments, and the question a developer asks on one duplicate never reaches the next developer who could answer it.

AI is making teams faster, and it's making output and likely ticket sprawl faster as well. So instead of asking AI to create more cards, I asked it to create fewer.

The setup is simple: an AI assistant, Favro's official MCP server, and four skills. A skill is a saved set of instructions that turns a one-off prompt into a repeatable workflow. I used Claude in the video, but any AI assistant that can connect to Favro's MCP server can run the same workflows. Each one shows its plan and waits for approval before it changes anything.

Problem: Teams copy the same work item onto every board that touches it. The copies drift apart, and nobody knows which one holds the latest notes and comments.

Solution: In Favro, one card can sit on several team boards at once, with one description and one comment thread. The AI finds the copies, summarizes every duplicate set, and asks set by set whether to merge. A merge puts one card on every board in each team's own column, keeps every copy's notes under a heading for that team, carries the comments over, and archives the copies.

Skill: favro-one-item-every-team

Bonus note for (former) JIRA users: If you use(d) the Favro-JIRA integration to migrate your JIRA issues to Favro cards, you end up with a pile (because that’s how JIRA works). That pile you can now consolidate using this skill.

Problem: Every feedback session (e.g. a sprint review or a playtest) produces a pile of notes, and the same issue gets filed again and again, often as a new ticket for something that already has one.

Solution: The AI reads the playtest summary, groups notes that describe the same issue, and searches Favro before creating anything. Issues that already have a card get the player quotes as one evidence comment. New issues become new cards on the right team's board, tagged with a severity. Praise gets reported but never filed.

Skill: favro-playtest-triage

Problem: QA files bugs on its own board, so the feature team never sees them associated with the feature they belong to.

Solution: I built this skill for a Favro customer. The AI reads the bugs linked to a feature and places each one under that feature on the feature backlog. The bug also stays on the Bug Board, so QA and the feature team share one card.

Skill: favro-link-bugs-to-features

Problem: Work stalls quietly. By the time anyone asks about a card that hasn't moved, a week or two is gone.

Solution: AI scans one board for cards that are started but haven't moved in a set number of days. It sorts them into Parked (waiting on a stated reason), Quiet (being discussed but not moving), and Silent, then drafts a short nudge for each. Nothing is posted until approved.

Skill: favro-stuck-work-scan

All four follow the same idea: one backlog item, one card, one conversation. They also share the same guardrails. Every skill shows its plan first and changes nothing until you say yes. None of them deletes a card (only archives duplicates). Each one does the legwork a producer would otherwise have to do manually.

The video walks through all four in a sample game studio workspace. Download the four skills, free, with a short guide to using them with your AI assistant of choice. Once downloaded, the skills are yours to use as is or adapt to your own process. They come with no guarantee, so give each one a trial run in a sandbox collection before pointing it at live work.

Check all this out in practice in this video.