ecommerce-competitor-analyzer

Files (26)

• .gitignore
• LICENSE
• README.md
• SKILL.md
• INSTALLATION.md · docs/
• SETUP.md · docs/
• .gitkeep · output/
• platforms.yaml
• analysis-prompt-base.md · prompts/
• n8n-workflow-analysis.md · references/
• .gitkeep · reports/
• auth-google-sheets-manual.js · scripts/
• auth-google-sheets-service-account.js · scripts/
• auth-google-sheets.js · scripts/
• auth-guide.sh · scripts/
• auth-manual.js · scripts/
• auth-oob.js · scripts/
• auth-simple.js · scripts/
• batch-processor.js · scripts/
• create-release.sh · scripts/
• exchange-token.js · scripts/
• google-sheets-writer.js · scripts/
• olostep-client.js · scripts/
• scrape-amazon.js · scripts/
• test-skill.js · scripts/
• verify-env.js · scripts/

SKILL.md (first lines)

# E-commerce Competitor Analyzer Skill

## Quick Start (For AI)

**When to use this skill**: When user asks to analyze, research, or extract insights from e-commerce products (Amazon, Temu, Shopee).

**What you should do**:
1. Extract product identifiers (ASINs or URLs) from user input
2. Call the scraper script to get product data
3. Call the AI analysis with the analysis prompt template
4. Output results in BOTH formats: Google Sheets + Markdown

**Input examples**:
- "Analyze B0C4YT8S6H"
- "Analyze these products: B0C4YT8S6H, B08N5WRQ1Y, B0CLFH7CCV"
- "Research this competitor: https://amazon.com/dp/B0C4YT8S6H"

**Output requirements**:
- Google Sheets table with: ASIN, Title, Price, Rating, 4 analysis summaries
- Markdown report with detailed 4-dimensional analysis

---

## How AI Should Process Requests

### Step 1: Extract Product Identifiers

From user input, extract all ASINs and/or URLs:

**Example inputs**:
```
"Analyze these Amazon products:
B0C4YT8S6H
B08N5WRQ1Y
B0CLFH7CCV"
```

**Extract**: `['B0C4YT8S6H', 'B08N5WRQ1Y', 'B0CLFH7CCV']`

**Mixed input handling**:
```
"Analyze B0C4YT8S6H and https://amazon.com/dp/B08N5WRQ1Y"
```

**Extract**: `['B0C4YT8S6H', 'B08N5WRQ1Y']` (extract ASIN from URL)

### Step 2: Batch Scrape Product Data

For each product identifier:
1. Detect platform (use `scripts/detect-platform.js` if available)
2. Call appropriate scraper (Amazon: `scripts/scrape-amazon.js`)
3. Use Olostep API with configured API key from `.env`

**Batch processing pattern**:
```javascript
// Process all products in parallel
const products = ['B0C4YT8S6H', 'B08N5WRQ1Y', 'B0CLFH7CCV'];
const results = await Promise.allSettled(
  products.map(asin => scrapeAmazon(asin))
);

// Handle failures gracefully
const successful = results.filter(r => r.status === 'fulfilled');
const failed = results.filter(r => r.status === 'rejected');
```

### Step 3: Batch AI Analysis

For each successfully scraped product:
1. Read the analysis prompt from `prompts/analysis-prompt-base.md`
2. Replace product data placeholders in the prompt
3. Call Gemini API (model: gemini-3-flash-preview)
4. Extract structured analysis results

**Analysis framework** (4 dimensions):
1. **文案构建逻辑与词频分析** (The Brain) - Copywriting strategy & keywords
2. **视觉资产设计思路** (The Face) - Visual design methodology
3. **评论定量与定性分析** (The Voice) - Review sentiment analysis
4. **市场维态与盲区扫描** (The Pulse) - Market positioning & blind spots

### Step 4: Generate Dual Format Output

**Format 1: Google Sheets** (Structured Data)

Write to Google Sheets with columns:
| ASIN | 产品标题 | 价格 | 评分 | 文案分析摘要 | 视觉分析摘要 | 评论分析摘要 | 市场分析摘要 |

**Sheet selection priority**:
1. User explicitly specified Sheet ID/Name/URL
2. Default from `.env` (`GOOGLE_SHEETS_ID`)
3. Ask user to provide Sheet ID

**Format 2: Markdown Report** (Detailed Analysis)

Generate file: `竞品分析-YYYY-MM-DD.md`

Structure:
```markdown
# Amazon Competitor Analysis Report

## Analysis Overview
- Products analyzed: 3
- Analysis date: 2026-01-29
- Total time: ~5 minutes

---

## Product 1: B0C4YT8S6H

### Basic Information
- Title: [Product title]
- Price: [Price]
- Rating: [Rating]

### Copywriting Strategy & Keyword Analysis
[Full analysis...]

### Visual Asset Design Methodology
[Full analysis...]

From aAAaqwq/AGI-Super-Team, MIT.

More from aAAaqwq/AGI-Super-Team