Review Analyst Agent

Review Analyst Agent

Analyze product reviews to find issues and prioritize improvements.

This skill uses 4 specialized agents that analyze reviews from different angles, then synthesizes into actionable recommendations.

What It Produces

| Output | Description | |--------|-------------| | Sentiment Overview | Overall sentiment breakdown (positive/neutral/negative) | | Top Complaints | Prioritized list of issues by frequency and severity | | Top Praise | What customers love (to protect/emphasize) | | Feature Requests | What customers want that doesn't exist | | Priority Matrix | Critical/Important/Nice-to-have improvements | | Action Plan | Specific recommendations with expected impact |

Prerequisites

• Web access for scraping reviews
• No API keys required

Workflow

Step 1: Identify Product and Sources (REQUIRED)

⚠️ DO NOT skip this step. Use interactive questioning — ask ONE question at a time.

Question Flow

⚠️ Use the AskUserQuestion tool for each question below. Do not just print questions in your response — use the tool to create interactive prompts with the options shown.

Q1: Product > "I'll analyze reviews for your product! First — what's the product? > > (Product name or URL)"

Wait for response.

Q2: Sources > "Where should I look for reviews? > > - Amazon > - App Store / Google Play > - G2 / Capterra > - Reddit > - All of the above > - Or specify"

Wait for response.

Q3: Context > "Is this your product or a competitor's? > > (Helps frame the analysis)"

Wait for response.

Q4: Issues > "Any known issues you want me to validate or explore? > > - Yes — describe them > - No — find all issues"

Wait for response.

Quick Reference

| Question | Determines | |----------|------------| | Product | What to analyze | | Sources | Where to scrape reviews | | Context | Framing of recommendations | | Issues | Focus areas for analysis |

---

Step 2: Collect Reviews

Use browser tools to scrape reviews from:

| Source Type | Platforms | |-------------|-----------| | E-commerce | Amazon, Walmart, Target, Best Buy | | Software | G2, Capterra, TrustRadius, Product Hunt | | Apps | App Store, Google Play Store | | General | Trustpilot, BBB, Yelp | | Social | Reddit, Twitter/X, YouTube comments | | Forums | Product-specific communities |

Collect for each review:

• Rating (if available)
• Date
• Review text
• Helpful votes (if available)

---

Step 3: Run Specialized Analysis Agents in Parallel

Deploy 4 agents, each analyzing from a different perspective:

Agent 1: Review Scraper

Focus: Find and collect reviews from multiple sources

Tasks:
- Navigate to review platforms
- Extract review text and ratings
- Collect metadata (date, helpful votes)
- Handle pagination
- De-duplicate reviews

Agent 2: Sentiment Analyzer

Focus: Analyze sentiment and emotional patterns

Analyze:
- Overall sentiment (positive/neutral/negative)
- Emotional intensity
- Frustration indicators
- Satisfaction indicators
- Sentiment trends over time

Agent 3: Issue Identifier

Focus: Categorize complaints and find patterns

Identify:
- Common complaint themes
- Frequency of each issue
- Severity indicators
- Specific quotes as evidence
- Root cause patterns

Agent 4: Improvement Recommender

Focus: Prioritize and recommend fixes

Recommend:
- Priority ranking of issues
- Specific improvement suggestions
- Expected impact of each fix
- Quick wins vs long-term investments
- Competitive gaps to address

---

Step 4: Synthesize into Analysis Report

Combine all agent outputs into a structured report:

{
  "product": {
    "name": "Product Name",
    "sources_analyzed": ["Amazon (342 reviews)", "Reddit (89 posts)", "G2 (56 reviews)"],
    "total_reviews": 487,
    "date_range": "Jan 2025 - Jan 2026",
    "analysis_date": "2026-01-04"
  },
  "sentiment": {
    "overall_score": 3.8,
    "breakdown": {
      "positive": 62,
      "neutral": 18,
      "negative": 20
    },
    "trend": "Improving (up from 3.5 six months ago)",
    "net_promoter_estimate": 32
  },
  "top_complaints": [
    {
      "rank": 1,
      "issue": "Battery drains too fast",
      "frequency": 47,
      "percentage": "23% of negative reviews",
      "severity": "High",
      "sample_quotes": [
        "Battery only lasts 2 hours, not the 8 advertised",
        "Have to charge it 3x per day",
        "Battery life is a dealbreaker"
      ],
      "root_cause": "Hardware limitation or software optimization needed",
      "recommendation": "Improve battery capacity or optimize power consumption",
      "expected_impact": "Could improve rating by 0.3-0.5 stars"
    },
    {
      "rank": 2,
      "issue": "App crashes frequently",
      "frequency": 32,
      "percentage": "16% of negative reviews",
      "severity": "High",
      "sample_quotes": [
        "App crashes every time I try to sync",
        "Lost all my data after app crashed"
      ],
      "root_cause": "Sync functionality stability",
      "recommendation": "Stability audit of mobile app, fix crash on sync",
      "expected_impact": "Could reduce 1-star reviews by 15%"
    }
  ],
  "top_praise": [
    {
      "feature": "Build quality",
      "frequency": 89,
      "percentage": "45% of positive reviews",
      "sample_quotes": [
        "Feels premium in hand",
        "Solid construction, very durable"
      ],
      "recommendation": "Emphasize in marketing, protect in future versions"
    }
  ],
  "feature_requests": [
    {
      "request": "Water resistance",
      "frequency": 23,
      "sample_quotes": [
        "Wish I could use it in the rain",
        "Would pay extra for waterproof version"
      ],
      "recommendation": "Consider for v2 or premium tier"
    }
  ],
  "competitor_mentions": [
    {
      "competitor": "Competitor X",
      "context": "Switching from",
      "frequency": 15,
      "sentiment": "Mixed - some prefer us, some prefer them"
    }
  ],
  "priority_matrix": {
    "critical": [
      {"issue": "Battery life", "reason": "Top complaint, high severity"},
      {"issue": "App crashes", "reason": "Causes data loss, drives 1-star reviews"}
    ],
    "important": [
      {"issue": "Water resistance", "reason": "Frequent request, competitive gap"}
    ],
    "nice_to_have": [
      {"issue": "Color options", "reason": "Low frequency, low impact"}
    ]
  },
  "action_plan": [
    {
      "priority": 1,
      "action": "Fix app crash on sync",
      "effort": "Medium",
      "impact": "High",
      "expected_outcome": "Reduce 1-star reviews by 15%"
    },
    {
      "priority": 2,
      "action": "Improve battery life or set realistic expectations",
      "effort": "High",
      "impact": "High",
      "expected_outcome": "Improve rating by 0.3-0.5 stars"
    },
    {
      "priority": 3,
      "action": "Add water resistance to roadmap for v2",
      "effort": "High",
      "impact": "Medium",
      "expected_outcome": "Address top feature request"
    }
  ]
}

---

Step 5: Deliver Actionable Insights

Delivery message:

"✅ Review analysis complete!

Product: [Name] Reviews Analyzed: [Count] from [Sources] Overall Sentiment: [Score] ([Positive]% positive)

Top 3 Issues (by frequency):

1. 🔴 [Issue 1] - [X]% of complaints
2. 🔴 [Issue 2] - [X]% of complaints
3. 🟡 [Issue 3] - [X]% of complaints

What Customers Love: ✅ [Praised feature 1] ✅ [Praised feature 2]

Priority Action: → Fix [Top Issue] first - expected to improve rating by [X]

Want me to:

• Deep dive on any issue?
• Compare to competitor reviews?
• Track changes over time?
• Create improvement roadmap?"

---

Integration with Other Agents

review-analyst-agent
    ↓ "Battery is top complaint"
product-engineer-agent
    ↓ "Design better battery solution"
patent-lawyer-agent
    ↓ "Check if solution is patentable"
copywriter-agent
    ↓ "Update marketing to address concern"

| Agent | How It Uses Review Data | |-------|-------------------------| | product-engineer-agent | Inform what to fix/improve | | competitive-intel-agent | Compare to competitor reviews | | market-researcher-agent | Validate market needs | | copywriter-agent | Address concerns in marketing | | pitch-deck-agent | Show customer-centric improvements | | media-utils | Generate PDF report from analysis |

---

Generate PDF Report

After completing the analysis, offer to generate a PDF:

> "Would you like me to generate a PDF report of this review analysis?"

python3 ${CLAUDE_PLUGIN_ROOT}/skills/media-utils/scripts/report_to_pdf.py \
  --input review_analysis.md \
  --output review_analysis.pdf \
  --title "Customer Review Analysis" \
  --style business

---

Agents

| Agent | File | Focus | |-------|------|-------| | Review Scraper | review-scraper.md | Find and collect reviews | | Sentiment Analyzer | sentiment-analyzer.md | Analyze sentiment patterns | | Issue Identifier | issue-identifier.md | Categorize complaints | | Improvement Recommender | improvement-recommender.md | Prioritize and recommend |

---

Example Prompts

Your product: > "Analyze reviews for our Bluetooth headphones on Amazon"

Competitor: > "What are people complaining about with Notion?"

Comparison: > "Compare reviews of our product vs Competitor X"

Feature focus: > "Find feature requests for our mobile app from App Store and Reddit"

Priority: > "What should we fix first based on customer feedback?"

Trend: > "How has sentiment changed over the last 6 months?"

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: michaelboeding
• Source: michaelboeding/skills
• License: MIT

Install and usage instructions live in the source repository linked above.