| name | prioritize |
| description | Rank open GitHub epics and issues by priority and recommend what to work on next. Read-only — never modifies issues. Invoke as /prioritize (optionally scoped to a label, milestone, or epic). |
| disable-model-invocation | true |
| argument-hint | ["optional label","milestone","or epic number"] |
Prioritize
Produce a prioritized view of the backlog. This skill is READ-ONLY — do not edit,
comment on, assign, or close anything.
1. Gather
Use gh to list open work. If $ARGUMENTS names a label, milestone, or epic, scope to
it; otherwise cover all open issues.
gh issue list --state open --limit 200 --json number,title,labels,assignees,milestone,createdAt,updatedAt
- For epics, also read child issues so blocked-by relationships are visible.
2. Score each item
Rate every issue on this rubric and show the factors, not just a number:
- Impact: user/business value or how much it unblocks.
- Effort: rough size (S/M/L) inferred from the description.
- Dependencies: is it blocked by, or blocking, other issues?
- Deadline: any milestone or date pressure.
- Staleness: time since last activity (use updatedAt).
Rank highest-leverage first: high impact, unblocks others, deadline-driven, low effort.
Flag anything blocked (and by what) and anything stale that's silently stuck.
3. Recommend
Output a ranked, skimmable table (rank, #, title, impact/effort, why) and a short
"START NEXT" line — the single epic or issue you'd pick up now and why. Assume I may be
reading this on my phone.