https://github.com/saqqdy/ai-code-reviewer-plus
AI-powered intelligent code reviewer â multi-dimensional semantic analysis, five-tier severity classification, actionable fix suggestions via Claude Code Skill
https://github.com/saqqdy/ai-code-reviewer-plus
ai-code-reviewer-plus claude-code claude-code-plugin claude-code-skill code-review intelligent-review pr-review semantic-analysis
Last synced: 6 days ago
JSON representation
AI-powered intelligent code reviewer â multi-dimensional semantic analysis, five-tier severity classification, actionable fix suggestions via Claude Code Skill
- Host: GitHub
- URL: https://github.com/saqqdy/ai-code-reviewer-plus
- Owner: saqqdy
- License: mit
- Created: 2026-06-21T08:36:21.000Z (26 days ago)
- Default Branch: master
- Last Pushed: 2026-07-04T07:54:48.000Z (13 days ago)
- Last Synced: 2026-07-04T09:14:06.795Z (13 days ago)
- Topics: ai-code-reviewer-plus, claude-code, claude-code-plugin, claude-code-skill, code-review, intelligent-review, pr-review, semantic-analysis
- Language: TypeScript
- Homepage: https://www.saqqdy.com/ai-code-reviewer-plus
- Size: 413 KB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
README
# đ AI Code Reviewer Plus
> AI-powered intelligent code reviewer â multi-dimensional semantic analysis, five-tier severity classification, actionable fix suggestions via Claude Code Skill.
[](https://www.npmjs.com/package/ai-code-reviewer-plus)
[](https://github.com/saqqdy/ai-code-reviewer-plus/blob/master/LICENSE)
[ä¸ćć楣](README_CN.md) | [ĺŽć´ć楣](https://ai-code-reviewer-plus.vercel.app)
---
## đŻ The Problem It Solves
Traditional linters catch syntax errors, but miss **semantic issues**:
```typescript
// Linter: â
No syntax errors
function processUser(input: any) {
return input.name.trim() // Runtime: đĽ if input is null
}
```
Questions linters can't answer:
- Is this business logic correct?
- Are there security vulnerabilities?
- Will this cause performance issues?
- Is this maintainable?
| Scenario | Traditional Linter | AI Code Reviewer |
|----------|-------------------|------------------|
| "Is this safe?" | Pattern matching, can't understand context | AI understands business semantics |
| "How to fix?" | Reports error only | Provides executable fix snippets |
| "Is it critical?" | Binary pass/fail | 5-tier severity: BLOCKER â SUGGESTION |
---
## ⨠The Solution
AI Code Reviewer uses **semantic understanding** across five dimensions:
| Dimension | Focus | Example Rules |
|-----------|-------|---------------|
| **Correctness** | Logic errors, null pointers | COR-001: Null check missing |
| **Security** | XSS, SQL injection, data leakage | SEC-001: User input unsanitized |
| **Performance** | N+1 queries, memory leaks | PER-001: Loop inside loop |
| **Maintainability** | Duplicate code, complexity | MAIN-001: Duplicated logic |
| **Best Practices** | Framework conventions | BP-001: Vue reactive rules |
## đ Five-Tier Severity
Not all issues are equal. AI Code Reviewer classifies by impact:
| Level | Icon | Meaning | Action |
|-------|------|---------|--------|
| **BLOCKER** | đŤ | Must fix, blocks merge | Fix immediately |
| **HIGH** | đ´ | Critical, breaks functionality | Fix this iteration |
| **MEDIUM** | đĄ | Important, affects quality | Plan to fix |
| **LOW** | đ˘ | Minor, polish level | Fix later |
| **SUGGESTION** | đĄ | Optional improvement | Optional |
---
## ⨠Core Features
### đ Multi-Dimensional Analysis
- **Correctness** â Logic errors, null pointers, boundary conditions
- **Security** â XSS, SQL injection, data leakage
- **Performance** â N+1 queries, memory leaks, unnecessary re-renders
- **Maintainability** â Duplicate code, complexity, naming
- **Best Practices** â Framework conventions, code style
### đ Five-Tier Severity Classification
| Level | Meaning | Action |
|-------|---------|--------|
| đŤ BLOCKER | Must fix, blocks merge | Fix immediately |
| đ´ HIGH | High priority | Fix this iteration |
| đĄ MEDIUM | Medium priority | Plan to fix |
| đ˘ LOW | Low priority | Fix later |
| đĄ SUGGESTION | Optional improvement | Optional |
---
## đ Getting Started
### Option 1: Claude Code Plugin (Recommended)
```bash
# In Claude Code, run:
/plugin marketplace add saqqdy/ai-code-reviewer-plus
/plugin install ai-code-reviewer-plus
```
#### Available Commands
| Command | Description |
|---------|-------------|
| `/review [branch]` | Review diff against target branch |
| `/review-file ` | Review a single file |
| `/review-commit ` | Review a specific commit |
### Option 2: CLI (Zero-Install)
```bash
npx ai-code-reviewer-plus review --branch main
npx ai-code-reviewer-plus review-file src/index.ts
npx ai-code-reviewer-plus review-commit abc1234
```
### Option 3: Programmatic Usage
```bash
pnpm add ai-code-reviewer-plus
```
```typescript
import { collectDiff, detectProject } from 'ai-code-reviewer-plus'
const diffs = await collectDiff({
root: process.cwd(),
targetBranch: 'main'
})
const project = await detectProject(process.cwd())
```
---
## đ Example Output
```
/review main
đ Reviewing diff against main...
đ Project: vue3 (typescript)
đ Files changed: 3
đ Review Summary:
- Total findings: 12
- đŤ Blockers: 1
- đ´ High: 3
- đĄ Medium: 5
- đ˘ Low: 2
- đĄ Suggestions: 1
đŤ Blockers (Must Fix):
- đŤ **SEC-001** [BLOCKER] `src/auth.ts:45`
**XSS vulnerability in user input**
User input is directly rendered without sanitization.
```typescript
element.innerHTML = userInput // đĽ Dangerous!
```
đĄ **Fix**: Sanitize user input before rendering.
```typescript
element.textContent = sanitize(userInput)
```
Confidence: đ˘ High
```
---
## đ Quick Experience
Not sure where to start? Follow this path:
| Step | Method | Time | What You'll See |
|------|--------|------|-----------------|
| 1 | CLI | 1 min | Diff collection + project detection |
| 2 | API example | 3 min | Programmatic interface + formatting |
| 3 | Skill | 5 min | **Full review** (rules + AI analysis) |
**Fastest path**: Install the Skill, run `/review main` â one command, full experience.
```bash
# Step 1: Try CLI instantly (in any git repo)
npx ai-code-reviewer-plus review --branch main
# Step 2: Run the API example
git clone https://github.com/saqqdy/ai-code-reviewer-plus.git
cd ai-code-reviewer-plus && pnpm install
npx tsx examples/basic-usage.ts
# Step 3: Full AI-powered review (in Claude Code)
/plugin install ai-code-reviewer-plus
/review main
```
---
## âď¸ Configuration
Create `.ai-code-reviewer-plus.yml` to customize:
```yaml
rules:
enabled:
- COR-001 # Null pointer check
- SEC-001 # XSS prevention
- SEC-002 # SQL injection
- PER-001 # N+1 query detection
disabled:
- STYLE-001 # Naming conventions
severityOverrides:
SEC-001: BLOCKER # XSS must fix immediately
excludePaths:
- node_modules/
- dist/
- coverage/
maxFindingsPerFile: 20
outputFormat: markdown
```
---
## đď¸ Project Structure
```
ai-code-reviewer-plus/
âââ .claude/skills/ai-code-reviewer-plus/ # Claude Code Skill
âââ src/ # TypeScript source
â âââ collectors/ # Git data collectors
â âââ utils/ # Utilities
â âââ types.ts # Type definitions
âââ docs/ # VitePress docs
âââ examples/ # Usage examples
```
---
## đ ď¸ Development
```bash
pnpm install # Install dependencies
pnpm run lint # ESLint check
pnpm run typecheck # TypeScript check
pnpm run test # Run tests
pnpm run build # Build (ESM + CJS)
pnpm run docs:dev # Start docs server
```
---
## đ Version Roadmap
| Version | Theme | Status |
|---------|-------|--------|
| v0.1.0 | Core framework + CLI + Skill | â
Current |
| v0.2.0 | Rule library + AI analysis engine | đ Planned |
| v1.0.0 | Production-ready + Marketplace | đ Planned |
---
## đ License
[MIT](./LICENSE)