Amazon Review Export

Amazon product review export and analysis agent. Extract, organize, and analyze Amazon reviews — export to structured format, identify sentiment patterns, su...

Security Scan

The skill is an instruction-only review analysis framework that mostly matches its description, but it advertises an 'export by ASIN' capability without providing code, credentials, or a clear, safe method to fetch Amazon reviews — this gap could lead the agent to attempt ad-hoc scraping via shell commands.

Purpose & Capability

Name/description: Amazon review export & analysis. What it actually requires: nothing (no env, no binaries, no code). The SKILL.md expects either pasted review text or an ASIN for 'review export', but no implementation for fetching reviews is provided. Asking for ASIN-based export without code or declared API credentials is a mismatch: either the user must paste data (which is fine) or the agent would need to scrape Amazon (not provided, potentially problematic).

Instruction Scope

The instructions are largely analytic frameworks and formatting guidelines and explicitly ask the user to paste review text or provide ASINs. They do not instruct the agent to read system files or request secrets. However, the command list (e.g., 'review export <asin>') is ambiguous about how exports occur; with allowed-tools: Bash, an agent could attempt network/shell actions to fetch reviews if it infers those steps — the doc does not constrain or describe those actions.

Install Mechanism

No install spec and no code files are present. This minimizes installation risk because nothing is written to disk by the skill itself.

Credentials

No environment variables, credentials, or config paths are requested. The skill does not ask for unrelated secrets, which is proportionate for an analysis-only tool assuming users supply review text manually.

Persistence & Privilege

Defaults (not always-included, user-invocable, model invocation allowed) — no elevated persistence or cross-skill config changes are requested.

What to consider before installing

This skill is primarily a guide and template for analyzing review text; it does not include code to fetch reviews by ASIN. Before installing or using it: (1) Decide whether you will paste reviews yourself — that is the intended safe usage. (2) If you expect automatic export from Amazon, ask the author how 'review export <asin>' is implemented and whether it uses an official API (and which credentials are required); automatic fetching likely requires additional code or scraping and may violate Amazon's terms. (3) Be cautious because allowed-tools: Bash means the agent could run shell commands to fetch data if not explicitly constrained — monitor any commands the agent proposes and avoid supplying AWS/Amazon credentials or other secrets. (4) If you need automated scraping, prefer a skill that documents its fetch mechanism and clearly declares required credentials or uses official APIs. Additional info that would change this assessment to benign: a clear implementation (code or documented API usage) for 'review export' that uses a legitimate API or an explicit statement that the skill operates only on user-pasted review text.

Amazon Review Export & Analyzer

Extract intelligence from Amazon product reviews — organize into structured data, analyze sentiment patterns, identify product improvement opportunities, and generate competitive insights from customer voice data.

Commands

review export <asin>              # structure reviews into exportable format
review analyze <reviews>          # full sentiment and pattern analysis
review sentiment <reviews>        # sentiment scoring breakdown
review patterns <reviews>         # find recurring themes and pain points
review compare <asin1> <asin2>    # compare review profiles between products
review insights <reviews>         # extract product improvement opportunities
review competitive <comp-reviews> # analyze competitor review weaknesses
review summary <reviews>          # executive summary of review data
review csv <reviews>              # format reviews as CSV-ready data
review report <asin>              # comprehensive review intelligence report

What Data to Provide

• Review text — paste reviews directly (as many as possible)
• Star rating distribution — number of reviews at each star level
• ASIN — product identifier
• Competitor reviews — for competitive analysis
• Time period — recent reviews vs. older reviews for trend analysis

Review Analysis Framework

Review Export Format

Structure raw reviews into:

Date,Rating,Title,Review Text,Verified,Helpful Votes,Reviewer
2024-01-15,5,"Great product","Very satisfied with...",Yes,12,Customer123
2024-01-10,2,"Disappointing","Expected better...",Yes,3,Customer456

Sentiment Analysis Framework

5-star rating interpretation:

⭐⭐⭐⭐⭐ (5-star): Delighted — read for what exceeds expectations
⭐⭐⭐⭐   (4-star): Satisfied — note any "but" qualifiers
⭐⭐⭐     (3-star): Neutral — mixed feelings, often most useful insights
⭐⭐       (2-star): Dissatisfied — specific complaints, high value for improvement
⭐         (1-star): Angry — often extreme cases, filter for systemic vs. one-off

Sentiment scoring:

Positive signals (+): "love", "perfect", "great", "amazing", "exactly what I needed"
Negative signals (-): "disappointed", "broke", "doesn't work", "waste", "returned"
Neutral signals (=): "okay", "fine", "average", "as expected", "decent"

Net Sentiment Score = (Positive reviews - Negative reviews) / Total reviews × 100
Target: Score > 60 = healthy product sentiment

Theme Identification (Qualitative Coding)

Categorize all reviews into themes:

Product quality themes:

□ Build quality / durability
□ Materials / finish quality
□ Sizing / dimensions (accurate vs. listing)
□ Performance (does it work as claimed?)
□ Longevity (how long does it last?)

Customer experience themes:

□ Packaging / unboxing experience
□ Instructions / ease of setup
□ Customer service experience
□ Shipping / delivery condition
□ Value for money perception

Use case themes:

□ Intended use (matches expected use case)
□ Alternative uses (how customers use it unexpectedly)
□ Gifting (bought as a gift)
□ Replacement (replacing specific previous product)
□ Professional vs. personal use

Frequency Analysis

Count mentions of each theme:

Theme                    Mentions    % of Reviews    Sentiment
Durable/sturdy           45          42%             Positive
Easy to assemble         38          35%             Positive
Instructions unclear     22          20%             Negative
Size smaller than shown  15          14%             Negative
Great value for money    52          48%             Positive

Priority fix threshold: Any negative theme appearing in >10% of reviews requires action.

Pain Point Extraction

From negative reviews, extract specific pain points:

Pain Point              Frequency   Severity    Fix Category
Product breaks quickly  23 mentions High        Product quality
Wrong size/dimensions   15 mentions Medium      Listing accuracy
No instructions         12 mentions Low         Packaging insert
Hard to clean           8 mentions  Low         Product design

Severity classification:

• High: Safety, complete product failure, cannot use product
• Medium: Significant disappointment, reduced usefulness
• Low: Minor inconvenience, still satisfied overall

Competitive Review Intelligence

From competitor reviews, extract:

Competitor weaknesses (from their negative reviews): → These are your differentiation opportunities

Competitor strengths (from their positive reviews): → Baseline expectations you must meet or exceed

Competitor Pain Points → Your Product Claims
"Instructions are confusing" → "Clear 10-step illustrated guide included"
"Flimsy material" → "Reinforced with aircraft-grade aluminum"
"Customer service ignores" → "24/7 support with 1-hour response guarantee"

Review Trend Analysis

Compare recent vs. older reviews:

Period          Avg Rating    Top Complaint        Top Praise
Last 90 days:   4.1           Size issues (18%)    Easy use (42%)
6-12 months:    4.4           No issues dominant   Quality (55%)
12+ months:     4.6           Rare complaints      Durability (60%)

Trend: Rating declining → investigate recent product/supplier change

VOC (Voice of Customer) Summary

Generate a customer perspective summary:

WHAT CUSTOMERS LOVE (keep and amplify in marketing):
1. [Most praised attribute + quote]
2. [Second most praised + quote]
3. [Third most praised + quote]

WHAT CUSTOMERS WANT IMPROVED (product/listing fixes):
1. [Top pain point + specific ask]
2. [Second pain point + ask]
3. [Third pain point + ask]

WHAT SURPRISES CUSTOMERS (unintended uses or unexpected positives):
1. [Unexpected use case]
2. [Unexpected benefit]

Review-to-Listing Optimization

Map review insights directly to listing improvements:

Review insight → Listing change
"Sturdy, holds 50lbs easily" → Add to bullets: "HEAVY-DUTY CONSTRUCTION — tested to hold up to 50 lbs"
"Works great as a gift" → Title: add "Perfect Gift" / create gift-focused image
"Instructions confusing" → Add instruction image to image gallery
"Looks exactly as shown" → Emphasize "true-to-photo" in listing

Workspace

Creates ~/review-data/ containing:

• exports/ — structured CSV exports per ASIN
• analyses/ — full review analysis reports
• themes/ — coded theme frequency data
• competitive/ — competitor review intelligence
• voc/ — voice of customer summaries

Output Format

Every review analysis outputs:

1. Rating Distribution — star breakdown with percentage for each level
2. Net Sentiment Score — overall sentiment health (0-100)
3. Top 5 Positive Themes — what customers love most (with frequency)
4. Top 5 Negative Themes — main pain points (with frequency + severity)
5. VOC Summary — customer voice in plain language
6. Listing Optimization Map — review insights → specific listing improvements
7. Product Development Signals — engineering/sourcing changes implied by feedback
8. CSV Export — structured data ready to paste into spreadsheet