summarizing-sentry-issues
GitHub基于Sentry数据生成问题摘要，按主题和影响力分类，支持晨间简报与可靠性查询。通过获取未解决Issue及Uptime状态，结合业务影响评分进行排序，最终输出结构化报告，辅助快速掌握系统稳定性状况。
Trigger Scenarios
Install
npx skills add Tracer-Cloud/opensre --skill summarizing-sentry-issues -g -y
SKILL.md
Frontmatter
{
"name": "summarizing-sentry-issues",
"tools": [
"search_sentry_issues",
"get_sentry_issue_details",
"list_sentry_issue_events",
"get_sentry_uptime_digest"
],
"description": "Summarise Sentry issues by theme and impact with optional uptime rollup. For overviews, morning digests, reliability questions."
}
Sentry Summary
Sentry-only (#3 on-demand, #10 morning digest). Multi-source: finish Sentry here, then suggest a multi-source investigation.
1. Fetch issues
search_sentry_issues with query: "is:unresolved":
- 24h — morning digest, overnight (#10); 7d — "this week", overview.
Max 100 groups per page. Use digest.page_saturated vs page_complete.
When completeness is empty, report zero — do not widen unless asked.
2. Fetch uptime
get_sentry_uptime_digest. Skip if uptime_watch_active is false. Include
still-down before Issues, recovered after Issues.
3. Classify
digest.structural_clusters → business themes with sample_short_ids.
4. Rank
digest.priority_candidates + business_impact_score, not raw count.
Prefer userCount, blockers, regressions; cite impact_reasons.
5. Enrich
Top 3–5 only: get_sentry_issue_details + list_sentry_issue_events (limit 10).
6. Summarise
- Header + summary bullet counts (Issues + Uptime).
[DOWN]above Issues; Issues clusters + priority table;[RECOVERED]after.
Traps
count= events/group;issue_count= groups returnedpage_saturated= first page;completeness: empty= zero groups- Uptime needs a running watch schedule — no watch = no data
Version History
- 4303874 Current 2026-09-22 10:48
