GitHub Issue analysis

Analyze repository Issue trends and produce a weekly report.

When to use it

Repositories where the Issue volume has passed "readable". It answers where problems are clustering this week — it does not triage individual issues for you.

Flow

Steps

1. Fetch — issues in range with comment counts, labels and state
2. Classify — combine existing labels with title keywords into problem domains
3. Count — opened, closed, net, and stale (no response for N days)
4. Spot outliers — a domain suddenly spiking, or close rate falling off
5. Report — as Markdown, or pushed to a chat

Options

Option	Description	Default
repos	Repositories, owner/name	required
range	Reporting window	7d
stale_days	Days without response to count as stale	14
group_by	Grouping: label | keyword | both	both

Output

{
  "range": "2026-08-05..2026-08-11",
  "opened": 47,
  "closed": 39,
  "net": 8,
  "stale": 23,
  "top_domains": [
    { "domain": "Install and environment", "count": 14, "trend": "+180%" },
    { "domain": "API compatibility", "count": 9, "trend": "-10%" }
  ]
}

Rate limits

Unauthenticated requests get 60 per hour, which a large repository exhausts almost immediately. Configure a token — authenticated requests get 5,000 per hour.