AgentSkillsCN

Prioritize

按严重程度对评审意见进行优先级排序

SKILL.md
--- frontmatter
description: Prioritize review comments by severity
model: sonnet
allowed-tools: Read

Prioritize Review Comments

Categorize and prioritize review comments for efficient resolution.

Purpose

Level 2 (Workflow)

Analyze review comments and categorize by severity to focus on critical issues first.

Variables

COMMENTS: $ARGUMENTS # Review comments (from /review/fetch output or direct input)

Severity Categories

🔴 MUST_FIX (Critical)

Issues that block merge. Fix these first.

Security vulnerabilities:

  • •SQL injection, XSS, CSRF
  • •Authentication/authorization bypass
  • •Secrets or credentials in code
  • •Insecure data handling

Logic errors:

  • •Incorrect algorithm implementation
  • •Race conditions, deadlocks
  • •Null pointer / undefined access
  • •Off-by-one errors
  • •Incorrect business logic

Data integrity:

  • •Data loss risks
  • •Missing validation
  • •Incorrect state management

Keywords to watch for:

  • •"security", "vulnerability", "injection", "auth"
  • •"bug", "error", "incorrect", "wrong"
  • •"crash", "null", "undefined", "race"
  • •"must", "required", "blocking"

🟡 SHOULD_FIX (Important)

Issues that should be addressed but don't block merge.

Performance issues:

  • •N+1 queries
  • •Memory leaks
  • •Blocking operations
  • •Inefficient algorithms

Best practices:

  • •Missing error handling
  • •Inadequate logging
  • •Missing tests for new code
  • •Unclear error messages

Maintainability:

  • •Complex/hard-to-read code
  • •Missing documentation for public APIs
  • •Tight coupling

Keywords to watch for:

  • •"performance", "slow", "inefficient"
  • •"should", "consider", "recommend"
  • •"test", "coverage", "logging"

🟢 OPTIONAL (Nice to Have)

Suggestions that improve quality but are discretionary.

Style issues:

  • •Naming conventions
  • •Code formatting
  • •Comment quality
  • •File organization

Info suggestions:

  • •Alternative approaches
  • •Minor optimizations
  • •Future considerations

Keywords to watch for:

  • •"nitpick", "nit", "minor"
  • •"suggestion", "optional", "could"
  • •"style", "formatting", "naming"

Workflow

1. Parse Comments

Extract each comment with context:

  • •File path and line number
  • •Author (human vs automated)
  • •Full comment text
  • •Any thread/replies

2. Classify Each Comment

For each comment, evaluate:

  1. •Check for security keywords → 🔴 MUST_FIX
  2. •Check for bug/error indicators → 🔴 MUST_FIX
  3. •Check for performance concerns → 🟡 SHOULD_FIX
  4. •Check for best practice suggestions → 🟡 SHOULD_FIX
  5. •Check for style/nitpick indicators → 🟢 OPTIONAL

Consider author weight:

  • •Copilot security warnings → take seriously
  • •Human reviewer "must fix" → 🔴
  • •Human reviewer suggestions → evaluate context

3. Sort by Priority

Order within each category by:

  1. •Severity of impact
  2. •Ease of fix (quick wins first)
  3. •File location (group related fixes)

Report

markdown
## Prioritized Review Comments

### Summary

| Priority | Count | Action |
|----------|-------|--------|
| 🔴 MUST_FIX | X | Required before merge |
| 🟡 SHOULD_FIX | X | Recommended |
| 🟢 OPTIONAL | X | At your discretion |

---

### 🔴 MUST_FIX (Critical) - X items

These must be addressed before merging.

#### 1. [file.py:42] Security: SQL injection risk
**Author:** @copilot
**Comment:**
> User input is passed directly to SQL query without sanitization.

**Suggested fix:** Use parameterized queries.

---

#### 2. [auth.py:15] Logic: Missing null check
**Author:** @reviewer
**Comment:**
> This will crash if user is not authenticated.

**Suggested fix:** Add null check before accessing user properties.

---

### 🟡 SHOULD_FIX (Important) - X items

These should be addressed if time permits.

#### 1. [api.py:88] Performance: Potential N+1 query
**Author:** @copilot
**Comment:**
> This query inside a loop could cause performance issues.

**Suggested fix:** Batch the queries or use eager loading.

---

### 🟢 OPTIONAL (Nice to Have) - X items

Address these at your discretion.

#### 1. [utils.py:23] Style: Consider renaming
**Author:** @reviewer
**Comment:**
> `data` is a bit vague, consider `user_records`.

---

## Action Plan

1. **Immediate:** Fix all 🔴 MUST_FIX items
2. **Before merge:** Address 🟡 SHOULD_FIX items
3. **Optional:** Consider 🟢 OPTIONAL improvements

Estimated effort: X items, ~Y minutes

Examples

Example 1: Prioritize from fetch output

code
/review/fetch | /review/prioritize

Example 2: Direct input

code
/review/prioritize "Security: SQL injection at line 42"

Integration Points

  • •Receives input from /review/fetch
  • •Output guides fix prioritization in /workflow/merge-cycle
  • •Helps decide which issues to auto-fix vs manual fix