bug-triage-prep（错误分类准备）

Bug Triage Prep

Turn a list of open bugs into clear triage notes with enough context to decide priority and next action.

Core rule

• Focus on the affected user workflow, repro quality, scope boundaries, and next decision.
• Keep business framing brief. The important thing is whether the bug is real, broad, severe, and actionable.
• Separate confirmed facts from assumptions.
• Do not just restate bug titles. Add triage judgment.

Use when

• The user needs triage prep for bug review.
• The team wants a prioritized defect summary.
• Open bugs need clearer notes before assignment or escalation.

Do not use when

• The user wants to create a new bug instead of triaging existing ones.
• The user needs a full status report across initiatives and tasks.
• There are no bugs to review.

Triage dimensions

Assess each bug using concrete signals:

• User impact: what the user cannot do or what breaks in the workflow.
• Customer reach: how many users, customers, or segments are likely affected.
• Repro reliability: always, intermittent, or unclear.
• Scope boundary: where it happens and where it does not.
• Workaround: whether users can still complete the job some other way.
• Evidence quality: logs, screenshots, support reports, or exact repro steps.

Workflow

1. Fetch bugs with list-bugs.
2. Use active statuses unless the user asks for a narrower slice.
3. Enrich bugs with get-task-details when the title or summary is too thin to triage responsibly.
4. If multiple bugs describe the same failure mode, call that out as a likely duplicate cluster.
5. Order the notes by triage urgency, not alphabetically.
6. Use the same note structure for every bug so the output is easy to scan and compare.

Output guidance

• Start with a short triage summary for the whole set.
• Then write one note per bug.
• Prefer ### headings for bug titles.
• Use markdown links for related work items when available.
• Keep each note compact but decision-ready.

Per-bug note format

### <bug title>

Short TLDR paragraph:
In 2-4 sentences, summarize what is broken, who is affected, how reliable the repro is, and why this bug should or should not be prioritized now.

### Background
- One short paragraph on the affected workflow and what is happening today.

### Repro and evidence
- Best known repro steps or trigger conditions.
- Evidence quality: confirmed, partial, or unclear.

### Known scope / boundaries
- Where this bug shows up.
- What appears unaffected or still unconfirmed.

### Impact and workaround
- What the user cannot do.
- Whether a workaround exists.

### Triage recommendation
- Suggested priority with a short reason.
- Suggested next action such as investigate, assign, merge with duplicate, wait for more evidence, or close.

### Open questions
- What is still missing to make a confident call?

Priority posture

Use direct language:

• Highest priority: blocks a core workflow, has broad reach, or has no workaround.
• High priority: serious user pain with solid evidence, but not a total blocker.
• Medium priority: real issue, narrower scope, partial workaround, or weaker evidence.
• Low priority: edge case, unclear repro, cosmetic issue, or low user impact.

If confidence is low, say that clearly instead of pretending the triage is settled.

Confidence rule

For every triage recommendation, be explicit about confidence:

• High confidence: strong repro or evidence, clear impact, little ambiguity.
• Medium confidence: likely real, but one of repro, reach, or scope is still fuzzy.
• Low confidence: weak evidence, unclear repro, or likely duplicate/noise.

Do not present a shaky recommendation as definitive.