ecommerce-design-patent-analyzer

name	ecommerce.design-patent-analyzer
version	1.0.0
description	Check visual/design patent similarity from product images. Triggers: design infringement, look-alike patent, visual similarity, design clearance. Use for appearance/design risk, not utility claims or legal-status checks.
allowed-tools	["Bash","Read","Write"]

IMPORTANT — Working Directory: All scripts in this skill MUST be executed with the working directory set to the absolute path of the directory containing this SKILL.md file (the skill root directory). Always cd into this directory before running any command. Do NOT use relative paths or shell shortcuts like ~ in paths, as they may not be resolved by the execution environment.

Design Patent Analyzer v1.0.0

Product Understanding Step

Before running this skill, if the user provides a product, image, ASIN, listing, product ID, or product idea, first identify:

• product type/category
• main use case
• key physical, design, or functional features
• target marketplace, country, or platform
• user's analysis goal

Use this product understanding to choose parameters, filters, competitors, regions, keywords, or analysis dimensions before executing the script.

Does this design have conflicts?

Design patent similarity analysis using dual-model image search against 170M+ patents.

Core Question

— Does this design have patent conflicts? — Is my product design safe to launch?

When to Use

• Before finalizing product design — check for conflicts early
• Before mold production — expensive to change later
• When copying competitor designs — know the risks
• For design registration — ensure novelty

Differs From / Not Applicable

• Use patent-risk-checker for broader product-level IP/FTO risk screening.
• Use patent-claim-analyzer for interpreting a known patent claim.
• Use patent-legal-status for validity/expiration questions.
• Use this skill when visual/design similarity is the main concern.

Workflow

1. Understand product appearance and target countries.
2. Run design/appearance patent similarity search from the image.
3. Rank similar designs by visual similarity and legal relevance.
4. Return risk level, closest matches, and next legal-review steps.

Data Sources

Source	Endpoint	Data
Zhihuiya	/api/v1/tools/research/zhihuiya/patentImageSearch (model=1)	Smart association — similar designs by appearance
Zhihuiya	/api/v1/tools/research/zhihuiya/patentImageSearch (model=2)	Exact match — shape and pattern-based matches
Zhihuiya	/api/v1/tools/research/zhihuiya/legalStatus	Legal status and litigation events

Coverage: 170M+ patents from 150+ countries. Full Locarno classification (LOC) support.

⚠️ MANDATORY: Charts & Display Rules

1. Generate charts by default for full analytical reports — use --chart with an output directory unless the user asks for a quick text-only result or chart data is unavailable.
2. Send generated charts to the user — when charts are created, share all chart PNGs immediately.
3. All chart styling is driven by ecommerce_chart_helpers.py which reads from chart_style.json (derived from references/display-rules.md). Do NOT hardcode colors in scripts.

Usage

# Basic search
python3 scripts/design_patent_analyzer.py '{"imageUrl": "https://example.com/product.jpg"}'

# With region filter
python3 scripts/design_patent_analyzer.py '{"imageUrl": "https://example.com/product.jpg", "countries": "US,CN,EU"}'

# Generate charts
python3 scripts/design_patent_analyzer.py '{"imageUrl": "..."}' --chart /tmp/charts

Parameters

Parameter	Type	Required	Default	Description
imageUrl	string	Yes	-	Product image URL
countries	string	No	all	Country filter (e.g., "US,CN,EU")
limit	int	No	30	Max patents per model
activeOnly	bool	No	true	Only active patents

Output Structure

The output will be a structured markdown report, following this format:

Product Design Patent Analysis Report

---

1. Executive Summary

• Analysis Date: [analysis_date]
• Data Source: [data_source]
• Input Image: [input.image_url] (Analyzed Countries: [input.countries], Active Patents Only: [input.active_only])
• Conflict Level: [conflict_summary.emoji] [conflict_summary.level] (Highest Similarity: [conflict_summary.max_similarity_pct])
• Core Recommendation: [conflict_summary.recommendation]
• Total Patents Found: [conflict_summary.total_patents_found]

2. Similarity Model Comparison

• Similar Patents Found by Different Models:

Model	Count Found	Total in Database	Highest Similarity
Smart Association	[model_comparison.smart_association.count]	[model_comparison.smart_association.total_in_db]	[model_comparison.smart_association.max_similarity]
Exact Match	[model_comparison.exact_match.count]	[model_comparison.exact_match.total_in_db]	[model_comparison.exact_match.max_similarity]

3. Conflict Summary

• Conflict Level Distribution:

Level	Count
Likely Conflict	[conflict_summary.conflict_counts.likely]
High Similarity	[conflict_summary.conflict_counts.high]
Potential Conflict	[conflict_summary.conflict_counts.potential]
None	[conflict_summary.conflict_counts.none]

4. Top Similar Patents Found

• List of Patents Ranked by Similarity (Showing Top 5):

Patent Number	Similarity	Title	Assignee	Filing Date	Status	Patent Image
[top_conflicts[0].patent_number]	[top_conflicts[0].similarity]	[top_conflicts[0].title]	[top_conflicts[0].assignee]	[top_conflicts[0].filing_date]	[top_conflicts[0].status]	[top_conflicts[0].image_url]
...	...	...	...	...	...	...

5. Locarno Classification Analysis

• Predicted Product Classification: [loc_analysis.predicted_product_loc]
• Total LOC Matches: [loc_analysis.total_loc_matches]
• Unique Classifications: [loc_analysis.unique_classifications]
• Top Classifications:
  • [loc_analysis.top_classifications[0].loc] (Count: [loc_analysis.top_classifications[0].count])
  • ...

6. Assignee Analysis

• Unique Assignees: [assignee_analysis.unique_assignees]
• Top Assignees (Showing Top 3):

Assignee Name	Patent Count	Highest Similarity
[assignee_analysis.top_assignees[0].name]	[assignee_analysis.top_assignees[0].patent_count]	[assignee_analysis.top_assignees[0].max_similarity]
...	...	...

7. Attached Visualizations

• Similarity Gauge (1_conflict_gauge.png)
• Model Comparison (2_model_comparison.png)
• Top Conflicts Distribution (3_top_conflicts.png)
• Assignee Map (4_assignees.png)
• Conflict Level Distribution (5_distribution.png)

Conflict Levels

Level	Similarity	Action
🟢 NO_CONFLICT	< 50%	Design appears novel — safe to proceed
🟡 POTENTIAL	50–70%	Review specific features, consider modifications
🟠 HIGH_SIMILARITY	70–85%	Significant redesign recommended
🔴 LIKELY_CONFLICT	≥ 85%	Abandon this design direction

LOC Classification

Locarno Classification (LOC) categorizes industrial designs:

Class	Category
01	Foodstuffs
02	Clothing
06	Furnishing
07	Household goods
09	Packaging
14	Recording equipment
21	Games & toys
26	Lighting

Charts Generated

Chart	Description	File
Similarity Gauge	Overall conflict level	1_conflict_gauge.png
Model Comparison	Smart vs Exact results	2_model_comparison.png
Top Conflicts	Similarity distribution	3_top_conflicts.png
Assignee Map	Top patent holders	4_assignees.png
Conflict Distribution	Level breakdown pie	5_distribution.png

Limitations

• Image quality affects search accuracy
• Some newer patents may not be indexed
• Results are advisory, not legal opinions
• Best for product appearance, not functional features

Environment Variables

Variable	Required	Description
NEXSCOPE_PROXY_BASE	Yes	NexScope proxy base URL
NEXSCOPE_API_KEY	Yes	NexScope proxy API key

References

• Locarno Classification: https://www.wipo.int/classifications/locarno/