AgentSkillsCN

issue-triage

使用 ISSUE_TRIAGE_GUIDE.MD 自动化问题分类。适用于分析 GitHub 问题时调用。根据分类公式为问题打分,分配优先级,建议车道划分,并映射至 Omi 层次。

SKILL.md
--- frontmatter
name: issue-triage
description: "Automate issue triage using ISSUE_TRIAGE_GUIDE.MD. Use when analyzing GitHub issues. Scores issues using triage formula, assigns priority levels, suggests lane assignment, and maps to Omi layers."

Issue Triage Skill

Automate GitHub issue triage using the Omi Issue Triage Guide.

When to Use

Use this skill when:

  • Analyzing GitHub issues
  • When user requests issue triage
  • When reviewing new issues
  • When prioritizing issues
  • When assigning issues to lanes

Capabilities

1. Score Issues

Calculate priority score using the triage formula from ISSUE_TRIAGE_GUIDE.MD:

Priority Score = (Core Layer Weight × Failure Severity) + Trust Impact + Frequency + Maintenance Leverage - Cost & Risk

2. Map to Omi Layers

Identify which primary layer the issue affects:

  • Capture (Weight: 5): Audio recording, device pairing, permissions, battery
  • Understand (Weight: 4): Speech-to-text, language detection, diarization
  • Memory (Weight: 4): Memory creation, syncing, storage, metadata
  • Intelligence (Weight: 3): Summaries, insights, action items
  • Retrieval / Action (Weight: 3): Search, asking Omi, tasks, exports
  • UX / Polish (Weight: 1): UI layout, animations, wording
  • Docs / Tooling (Weight: 1): Documentation, examples, tooling

3. Evaluate Scoring Factors

Assess each factor (1-5 scale):

Failure Severity:

  • 5: Completely broken
  • 4: Frequently fails
  • 3: Partially degraded
  • 2: Minor annoyance
  • 1: Cosmetic

Trust Impact:

  • 5: Data loss or missing memories
  • 4: Incorrect or corrupted memories
  • 3: Inconsistent behavior
  • 2: Confusing but recoverable
  • 1: No trust impact

Frequency:

  • 5: Happens daily
  • 4: Weekly
  • 3: Regular but situational
  • 2: Rare
  • 1: Edge case

Maintenance Leverage:

  • 5: Eliminates a class of bugs
  • 4: Improves observability or stability
  • 3: Neutral
  • 2: Adds complexity
  • 1: Increases long-term maintenance burden

Cost & Risk (subtracted):

  • 5: Cross-device + backend + firmware
  • 4: Core pipeline change
  • 3: Moderate
  • 2: Small
  • 1: Trivial

4. Assign Priority Levels

Based on score:

  • >= 30: P0 - Existential / must fix immediately
  • 22-29: P1 - Critical
  • 14-21: P2 - Important
  • < 14: P3 - Backlog

5. Suggest Lane Assignment

Assign to appropriate lane:

  • Maintainer Now: High-risk, cross-system, or architectural changes
  • Community Ready: Clear scope, safe changes, suitable for contributors
  • Needs Info: Missing repro steps, logs, versions, or clarity
  • Park: Out of scope or low leverage

Triage Rules

Follow these principles:

  • Issues are signals, not commands
  • Popularity does not determine urgency
  • Data loss outranks feature requests
  • Capture failures outrank intelligence improvements
  • Memory-first principle: If Omi fails to capture or preserve memory, nothing else matters

Workflow

  1. Read Issue: Analyze issue description, labels, comments
  2. Map to Layer: Identify primary Omi layer affected
  3. Evaluate Factors: Score each factor (1-5)
  4. Calculate Score: Apply triage formula
  5. Assign Priority: Map score to priority level (P0-P3)
  6. Suggest Lane: Recommend lane assignment
  7. Report: Provide triage summary with reasoning

Example Triage

Issue: Recording stops unexpectedly

Analysis:

  • Layer: Capture (5)
  • Severity: 5 (Completely broken)
  • Trust Impact: 5 (Data loss - missing recordings)
  • Frequency: 4 (Weekly)
  • Leverage: 4 (Improves stability)
  • Cost: 3 (Moderate)

Score: (5 × 5) + 5 + 4 + 4 - 3 = 35 → P0

Lane: Maintainer Now (high-risk, affects core functionality)

Related Resources

Documentation

  • ISSUE_TRIAGE_GUIDE.MD - Complete triage guide and formula

Rules

  • .cursor/rules/omi-specific-patterns.mdc - Omi architecture and priorities

Commands

  • /auto-triage - Automatically triage an issue