voc-amazon-reviews
Health Uyari
- No license — Repository has no license file
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 17 GitHub stars
Code Uyari
- network request — Outbound network request in mcp_server/server.py
Permissions Gecti
- Permissions — No dangerous permissions requested
Bu listing icin henuz AI raporu yok.
MCP server + Agent Skill for Amazon review intelligence. The data layer is the moat — backed by Shulex VOC OpenAPI (industry's most stable Amazon review feed).
VOC Amazon Reviews
The most stable, most complete Amazon review data — wired into any AI tool via MCP.
API-grade. 10 marketplaces. Up to 1,000 reviews per ASIN. No scraper fragility. Open source.
Quick Start • Demo • MCP Server • Usage • How It Works • Roadmap
Why this exists — the data is the moat
Most Amazon seller tools have a dirty secret: the data is the problem, not the AI. Scrapers break the day Amazon ships a CSS change. "Free" tools cap you at 10 reviews. Paid SaaS dashboards lock the raw data behind a UI you can't query from your own scripts. Multi-market support is afterthought. Updates are daily snapshots, not live.
We built this project around the opposite premise: the review data layer should be boring, reliable, complete, and AI-native. Everything else — sentiment analysis, pain-point extraction, listing copy generation, the MCP server — is downstream of that.
| Typical seller-tool data layer | This project | |
|---|---|---|
| Source | Web scraper or undocumented scraping API | Paid OpenAPI (Shulex VOC) |
| Reliability | Breaks when Amazon changes HTML | API-grade, no DOM dependencies |
| Markets | Usually US-only | 10: US, CA, MX, GB, DE, FR, IT, ES, JP, AU |
| Volume | 10–50 reviews per ASIN (free tier cap) | Up to 1,000 per ASIN per call |
| Freshness | Daily snapshots, sometimes cached for days | Live pull |
| Schema | Strings only — verified/helpful/variant often dropped | Full schema: verified-purchase, helpful votes, vine voice, variant, dates |
| Non-English markets | Often broken or omitted | Native — JP/DE/FR/IT/ES reviews captured verbatim, then AI-translated for analysis |
| Access pattern | Locked behind their UI | curl + JSON, fully scriptable, MCP-ready |
You can build any analysis layer on top. The data underneath won't be the bottleneck.
🆕 MCP Server — feed this data layer directly to Claude / Cursor / Windsurf
Claude / Cursor / Windsurf
│
│ MCP (stdio)
▼
mcp_server/ ← 4 tools, all powered by the data layer above
│
├── fetch_reviews ← the data, raw
├── analyze_reviews ← data → insights
├── voc_full ← one-shot
└── extract_listing_improvements ← data → copyable listing edits
Self-hosted, BYO API key, open source. See mcp_server/ for the install + Claude Desktop / Code config snippets.
Demo
Input an ASIN. Get deep bilingual insights in 5 seconds. Fetches real Amazon reviews via Shulex VOC API, then runs AI semantic analysis — sentiment, pain points, selling points, and listing optimization tips. Not keyword counting. Actual language understanding.
Features
Data layer (the foundation):
| Feature | Description |
|---|---|
| API-grade reliability | Shulex VOC OpenAPI — no scrapers, no CAPTCHAs, no IP bans |
| 10 marketplaces | US, CA, MX, GB, DE, FR, IT, ES, JP, AU — full coverage |
| Up to 1,000 reviews / ASIN | Configurable per call; most tools cap at 10–50 |
| Full review schema | Rating, body, verified-purchase, helpful votes, vine voice, variant, dates |
| Live, not cached | Every call hits the live API — no stale daily snapshots |
| Zero browser deps | curl + python3 only. No Selenium, no proxy farm, no Docker |
Analysis layer (built on top):
| Feature | Description |
|---|---|
| Sentiment Analysis | Positive / neutral / negative breakdown with percentages |
| Pain Points | Top 5 customer complaints with real quotes and mention counts |
| Selling Points | Top 5 things buyers love with real quotes and mention counts |
| Listing Optimization | Actionable copy suggestions backed by review data |
| Bilingual Output | Every insight in both English and Chinese |
| AI-native via MCP | Plug into Claude Desktop / Code / Cursor / Windsurf in one config block |
| Free to Start | 8 reviews = 5 credits. New accounts include starter credits |
Quick Start
Step 1 — Get your free API key (30 seconds):
Step 2 — Clone and run:
git clone https://github.com/mguozhen/voc-amazon-reviews.git
cd voc-amazon-reviews
export VOC_API_KEY="your-key"
bash voc.sh B08N5WRWNW
That's it. No Docker. No npm install. No config files.
Usage
# Quick analysis — 8 reviews (5 credits)
bash voc.sh B08N5WRWNW
# Deep analysis — 100 reviews (50 credits)
bash voc.sh B08N5WRWNW --limit 100
# Japan marketplace
bash voc.sh B08N5WRWNW --market JP
# Save report to file
bash voc.sh B08N5WRWNW --limit 100 --output report.md
Options
| Flag | Default | Description |
|---|---|---|
--limit N |
8 | Number of reviews to fetch |
--market CODE |
US | Amazon marketplace (US CA MX GB DE FR IT ES JP AU) |
--output FILE |
stdout | Save report to markdown file |
--help |
— | Show help |
Cost Guide
| Reviews | Credits | Use Case |
|---|---|---|
| 8 (default) | 5 | Quick competitor check |
| 50 | 25 | Product validation |
| 100 | 50 | Deep analysis |
| 200 | 100 | Comprehensive audit |
Sample Output
╔══════════════════════════════════════════════════════════════╗
║ VOC AI 分析报告 / VOC AI Analysis Report ║
║ ASIN: B099Z93WD9 | analyzed: 8 reviews ║
║ Market: US | Generated: 2026-04-19 ║
╚══════════════════════════════════════════════════════════════╝
📊 情感分布 / Sentiment Distribution
─────────────────────────────────────────
正面 Positive ████████░░░░░░░░░░░░ 37%
中性 Neutral ██░░░░░░░░░░░░░░░░░░ 13%
负面 Negative ██████████░░░░░░░░░░ 50%
🔴 Top 5 痛点 / Pain Points
═══════════════════════════════════════════════════════════════
1. 充电口故障 / Charging port moisture glitch(2 条提及)
「充电口提示有水分,已知bug,一周都没恢复」
"Moisture in charging port — known glitch, can't charge"
2. 视频卡顿 / Video stalling and weak connection(2 条提及)
「看视频经常卡顿暂停,给小孩看的时候很烦」
"Stalls out, pausing videos, really annoying"
🟢 Top 5 卖点 / Selling Points
═══════════════════════════════════════════════════════════════
1. 性价比高 / Great value for money(3 条提及)
「价格实惠,功能齐全,看电影看书玩游戏都行」
"Budget friendly, entertainment on the go"
2. 便携尺寸 / Perfect portable size(2 条提及)
「尺寸刚好,放包里轻松带上飞机看电影」
"Perfect size, light and easy to fit in my purse"
💡 Listing 优化建议 / Optimization Suggestions
═══════════════════════════════════════════════════════════════
1. Highlight budget-friendly and portability in title
2. Add charging port care instructions in A+ Content
3. Guide users to sideload popular apps
How It Works
┌─────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Input ASIN │────▶│ Shulex VOC API │────▶│ AI Analysis │
│ │ │ │ │ │
│ B099Z93WD9 │ │ • Real reviews │ │ • Sentiment │
│ --market US│ │ • 10 markets │ │ • Pain points │
│ --limit 8 │ │ • 5s response │ │ • Sell points │
│ │ │ • No scraping │ │ • Optimization │
└─────────────┘ └──────────────────┘ └────────┬────────┘
│
▼
┌─────────────────┐
│ Bilingual Report │
│ (EN + ZH) │
└─────────────────┘
File Structure
voc-amazon-reviews/
├── SKILL.md # Skill definition (Claude/OpenClaw)
├── voc.sh # Main entry point
├── fetch.sh # Shulex VOC API client
├── analyze.sh # AI analysis + report renderer
├── scraper.sh # Legacy browser scraper (deprecated)
├── tests/
│ ├── test_unit.sh # 50 unit tests
│ └── test_regression.sh # 17 regression tests (live API)
├── demo/
│ ├── voc-demo.gif # Demo recording
│ └── demo.sh # Demo script
└── docs/
├── GTM.md # Go-to-market strategy
├── ROADMAP.md # Product roadmap
└── STORY.md # Project narrative
FAQ
Where does the review data come from?Reviews are fetched via the Shulex VOC API — a legitimate data provider with proper Amazon data licensing. No scraping. No browser automation. Just a clean API call.
How much does it cost?| Component | Cost |
|---|---|
| VOC API (8 reviews) | 5 credits |
| VOC API (100 reviews) | 50 credits |
| AI analysis | Depends on your OpenClaw model |
New accounts include starter credits — enough for multiple analyses. Get your free API key.
Is this against Amazon's Terms of Service?No. This tool uses the Shulex VOC API, which is a licensed data provider. It does not scrape Amazon directly.
What about API keys and security?VOC_API_KEY is read from environment variables — never written to disk or printed to stdout.
Related
- VOC AI — Full-featured Amazon review analytics platform
- VOC Open API — Get your free API key
- Social Reply Bot — AI-powered social media bot
- Solvea — AI receptionist for small businesses
License
MIT
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi