Daily GitHub issue triage
Each weekday, triages new, stale, and needs-triage issues across the repos you pick — grouped, prioritized, with a suggested label and owner per issue.
Weekdays at 9:00 AMEngineering Emailed to you
Use this routine →Opens prefilled — review and save.
Works with
The instructions
Every weekday at 9am, triage the open-issue backlog across the GitHub repos I list and email me a grouped, prioritized triage list. Use the **github** connection — read-only by default; only suggest. Repos to triage: - owner/repo-one - owner/repo-two - <add the repos your team actually owns> For each repo, build three buckets using issue listing and cross-repo issue search: 1. **New in last 24h** — open issues created in the last 24 hours. Flag any that are unlabeled (no labels) and any that are unassigned — these need attention first. 2. **Gone stale** — open issues with no update in ~30 days (last updated before 30 days ago). These are candidates to close, ping, or re-prioritize. 3. **Needs triage / bugs** — open issues currently labeled `needs-triage` or `bug`. For each issue, capture: number, title, author, age (created/last-updated), current labels and assignees, and the issue's URL — that link is what I'll click, so include it for every issue. Then add judgment on top of the data: - For each **new** issue, suggest one label chosen **from that repo's existing label list** (read the repo's labels first — never invent a label), and suggest a *likely* owner. Frame both as suggestions to review, not decisions — they're inferences and may be wrong. - Group by repo, then by bucket, most urgent first (unlabeled/unassigned new issues, then bugs, then stale). Stay read-only: just list and suggest. Do **not** apply labels or assignees unless I've explicitly opted in above AND my connection has write access. If a repo returns no open issues, or a bucket is empty, say so plainly for that repo instead of padding the list — and never guess at issues you couldn't read. If the github connection can't reach a repo, name it and move on.
What a run emails you
GitHub triage — 30 Jun 2026 18 open issues need eyes: 4 new (2 unlabeled), 3 bugs, 11 stale. Read-only — suggestions below. acme/web New (last 24h) • #842 "Login redirect loops on Safari" — @dpark · unlabeled, unassigned → suggest label `bug` · likely owner @rivera (auth area) — review • #841 "Docs: env var name is wrong" — @ktan · unassigned → suggest label `documentation` · likely owner @docs-team — review Bugs (needs attention) • #815 "CSV export truncates >10k rows" — labeled `bug` · @lee · 6 days old Stale (no activity 30d+) • #602 "Dark mode flicker" — last update 38 days ago · @rivera acme/api New (last 24h): none. Stale: #311 "Retry backoff too aggressive" — 44 days quiet, no assignee. Nothing applied — these are suggestions to review.
How it works
- Attach your github connection — the routine reads open issues across the repos you list (issue listing + cross-repo issue search) and reads each repo's existing labels. Building the triage list needs only read access; applying anything is opt-in and off by default.
- It computes the buckets for real — new in 24h, unlabeled/unassigned, no activity in ~30 days, and currently labeled `needs-triage`/`bug` all come from issue search qualifiers and created/updated timestamps, not guesswork — then groups and prioritizes them and emails one linked list.
- The suggested label (picked from your repo's own labels) and likely owner are model inferences layered on the data, so they ship as suggestions to review — not as changes made on your behalf.
Make it yours
- Edit the repo list and the buckets — change the 24-hour window, the ~30-day stale threshold, or the labels it watches for (`needs-triage`, `bug`, plus your own like `P0` or `security`).
- Filter what you see — limit to one milestone, a single label, or only unassigned issues, so the email matches how your team actually triages.
- Opt in to auto-apply: if your github connection has write access, allow it to apply each suggested label automatically (still leave owner assignment to you) — otherwise it stays read-only and suggestion-only.
Your connections stay yours: tokens are encrypted at rest, and the agent only uses the GitHub access you grant — it reads issues and labels to build the list, and changes nothing unless you explicitly turn on apply.
Run “Daily GitHub issue triage” on autopilot
Copy it to your account, tweak the details, and it runs weekdays at 9:00 am.
Use this routine →More Engineering templates
GitHub releases to a sheetEngineering
Logs each repo's weekly releases as rows in a Google Sheet, so nothing slips.
Mondays at 9:00 AM
Weekly GitHub release digestEngineering
A Monday-morning summary of every new release across the repos and dependencies you care about.
Mondays at 9:00 AM
Weekly Cloudflare docs & feature digestEngineering
Every Friday, scan Cloudflare's developer docs for the Workers, Pages, and security features that matter to your setup — plus recent changelog posts — emailed as a prioritized digest.
Fridays at 9:00 AM
Weekly library version & migration digestEngineering
Every Monday, check Context7 for the current documented version, API, and migration notes of the libraries your project depends on — emailed as a short upgrade to-do list.
Mondays at 8:00 AM