← Back to Skills Marketplace
Eureka Patent Image Search
by
linkfox-ai
· GitHub ↗
· v1.0.0
· MIT-0
61
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install linkfox-eureka-patent-image-search
Description
通过Eureka专利平台进行专利图像检索（以图搜图），上传一张图片URL即可检索外观设计或实用新型的相似专利。当用户提到专利图像检索、以图搜专利、外观设计搜索、图片相似专利、设计专利检索、图像搜索专利、专利视觉搜索、patent image search, search patent by image, desi...
README (SKILL.md)
Eureka Patent Image Search (Single Image)
This skill guides you on how to perform patent image search (search-by-image) via the Eureka patent platform. Given a single image URL, it finds visually similar patents in the design or utility patent databases, supporting multiple search models and extensive filtering options.
Core Concepts
The Eureka Patent Image Search tool enables visual similarity search across patent databases:
-
Search Models — Four models are available depending on the patent type and search intent:
- Model 1: Design smart association — finds design patents with intelligent visual association
- Model 2: Design search this image — exact visual match for design patents
- Model 3: Utility match shape — matches utility patents by shape/contour
- Model 4: Utility match shape + pattern + color — matches utility patents by shape, pattern, and color combined
-
Patent Types — Two patent types are supported:
D(Design) — search within design/industrial design patents (use models 1 or 2)U(Utility) — search within utility model patents (use models 3 or 4)
-
Filtering — Extensive optional filters for country/authority, Locarno classification, date ranges, legal status, assignees, and keyword fields.
-
Scoring — Results include a similarity
scoreand can be sorted by score, application date, publication date, or issue date.
Parameter Guide
Required Parameters
| Parameter | Type | Description |
|---|---|---|
| url | string | The URL of the image to search with. Must be a publicly accessible image URL. |
| model | integer | Search model: 1 = design smart association, 2 = design search this image, 3 = utility match shape, 4 = utility match shape + pattern + color |
| patentType | string | Patent type to search: "D" = design patent, "U" = utility patent |
Optional Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| country | string | — | Patent authority codes, comma-separated (e.g., CN,US,JP,EP,WO) |
| loc | string | — | Locarno classification code. Supports AND/OR/NOT boolean operators. |
| applyStartTime | string | — | Application date range start (format: yyyyMMdd) |
| applyEndTime | string | — | Application date range end (format: yyyyMMdd) |
| publicStartTime | string | — | Publication date range start (format: yyyyMMdd) |
| publicEndTime | string | — | Publication date range end (format: yyyyMMdd) |
| mainField | string | — | Search keyword within specific fields: title, abstract, claims, description, pn, applicant, inventor, IPC, UPC, LOC |
| assignees | string | — | Filter by assignee/rights holder name |
| legalStatus | string | — | Legal status codes, comma-separated. Values: 1=Published, 2=Examining, 3=Granted, 11=Withdrawn, 13=Rejected, 14=Revoked, 15=Expired, etc. |
| simpleLegalStatus | string | — | Simplified legal status: 0=Invalid, 1=Valid, 2=Pending, 220=PCT expired, 221=PCT in period, 999=Undetermined |
| preFilter | integer | 1 | Enable (1) or disable (0) pre-filtering by country and LOC classification |
| scoreExpansion | boolean | — | Enable score expansion for broader results |
| stemming | integer | 0 | Enable (1) or disable (0) word stemming in keyword search |
| includeMachineTranslation | boolean | — | Include machine-translated content in search |
| field | string | SCORE | Sort field: SCORE (similarity), APD (application date), PBD (publication date), ISD (issue date) |
| order | string | desc | Sort order: desc or asc |
| limit | integer | 10 | Results per page, range 1–100 |
| offset | integer | 0 | Pagination offset, range 0–1000 |
| lang | string | — | Title language preference: original, cn, en |
| isHttps | integer | — | Return image URLs with HTTPS (1) or HTTP (0) |
| returnImgId | boolean | — | Include image ID in results |
Model & Patent Type Combinations
| Patent Type | Recommended Models | Description |
|---|---|---|
| D (Design) | 1, 2 | Model 1 for smart association, Model 2 for exact visual match |
| U (Utility) | 3, 4 | Model 3 for shape match, Model 4 for shape + pattern + color |
Response Fields
| Field | Description |
|---|---|
| patentId | The patent's internal ID |
| patentPn | Publication number |
| title | Patent title |
| url | URL of the similar image found in the patent |
| score | Similarity score |
| apdt | Application date |
| pbdt | Publication date |
| authority | Patent authority (country code) |
| inventor | Inventor name |
| apno | Application number |
| originalAssignee | Original applicant/assignee |
| currentAssignee | Current rights holder |
| loc | Array of Locarno classification codes |
| imgId | Image ID (when returnImgId is enabled) |
| locMatch | Whether the Locarno classification matched the filter |
| total | Number of results in current page |
| allRecordsCount | Total number of matching results |
| costToken | Token cost for this query |
Usage Examples
1. Search for similar design patents using an image
Search for design patents similar to this image: https://example.com/product-image.jpg — look in CN and US patent databases.
→ Use model=1 or 2, patentType="D", country="CN,US"
2. Find utility patents with a similar shape
Find utility patents that have a similar shape to this product image: https://example.com/gadget.png
→ Use model=3, patentType="U"
3. Search with date and legal status filters
Search for valid design patents similar to this image, filed after 2020, in the CN database: https://example.com/design.jpg
→ Use model=2, patentType="D", country="CN", applyStartTime="20200101", simpleLegalStatus="1"
4. Search with Locarno classification
Find design patents in LOC 14-01 similar to this product: https://example.com/phone-case.jpg
→ Use model=1, patentType="D", loc="14-01"
5. Search with keyword and assignee filters
Search for design patents similar to this image from Apple Inc.: https://example.com/device.jpg
→ Use model=2, patentType="D", assignees="Apple"
6. Get more results with pagination
Show me the next 20 design patents similar to this image: https://example.com/product.jpg
→ Use limit=20, offset=20
Display Rules
- Present results as a visual gallery when possible: Show the patent publication number, title, similarity score, and image URL for each result.
- Score interpretation: Higher scores indicate greater visual similarity. Present scores as percentages or relative rankings for user clarity.
- Key metadata: Always include patent number, title, authority, application date, and assignee for each result.
- Total count: Mention
allRecordsCountto indicate how many total matches were found, and how many are displayed. - Model explanation: When presenting results, briefly note which search model was used so the user understands the matching approach.
- Error handling: If the query fails, check common issues: invalid image URL, unsupported image format, invalid model/patentType combination.
- Pagination guidance: If there are more results than displayed, inform the user they can request more.
Important Limitations
- Single image only: This endpoint accepts one image URL per request.
- Image URL must be publicly accessible: Private or authentication-required image URLs will fail.
- Model & patent type pairing: Models 1–2 are for design patents (D), models 3–4 are for utility patents (U). Mismatched combinations may produce poor results.
- Offset limit: Maximum offset is 1000, meaning at most 1000 + limit results can be accessed via pagination.
- Locarno classification: The
locparameter uses Locarno codes; users may need help identifying the correct code for their product category.
User Expression & Scenario Quick Reference
Applicable — Visual patent search queries:
| User Says | Scenario |
|---|---|
| "Search for patents similar to this image" | Basic image search |
| "Find design patents that look like this" | Design patent visual search |
| "Are there utility patents with this shape" | Utility patent shape match |
| "Search CN and US design patents for this product image" | Filtered image search |
| "Find valid patents similar to this design" | Legal status-filtered search |
| "Show me more results" | Pagination request |
| "Search in Locarno class 14-01" | Classification-filtered search |
Not applicable — Needs beyond image search:
- Patent text-based search (keyword, classification, applicant)
- Patent bibliography or metadata retrieval
- Patent legal status queries
- Patent claims or description retrieval
- Patent family analysis
- Multi-image or batch image search
Feedback:
Auto-detect and report feedback via the Feedback API when any of the following apply:
- The functionality or purpose described in this skill does not match actual behavior
- The skill's results do not match the user's intent
- The user expresses dissatisfaction or praise about this skill
- Anything you believe could be improved
Call the feedback API as specified in references/api.md. Do not interrupt the user's flow.
For more high-quality, professional cross-border e-commerce skills, visit LinkFox Skills.
Usage Guidance
Before installing or enabling this skill: (1) be aware it sends the image URL and an API key to external hosts (tool-gateway.linkfox.com and skill-api.linkfox.com); do not use private/confidential images unless you trust those services. (2) The package metadata omits the required LINKFOXAGENT_API_KEY — confirm with the publisher and update the registry before providing credentials. (3) Verify who operates the LinkFox endpoints (homepage/maintainer identity, privacy policy, and API key scope) so the API key cannot be abused. (4) Prefer least-privilege API keys and avoid reusing high-privilege credentials. (5) If you need stronger assurance, ask the publisher for a homepage, source provenance, and a signed/registered publisher identity; without that, treat the skill as unvetted.
Capability Analysis
Type: OpenClaw Skill
Name: linkfox-eureka-patent-image-search
Version: 1.0.0
The skill bundle provides a legitimate tool for patent image searching via the Eureka platform. The Python script (eureka_patent_image_search.py) uses standard libraries to perform authenticated POST requests to a specified API endpoint (tool-gateway.linkfox.com) and handles parameters safely. The instructions in SKILL.md and references/api.md are consistent with the tool's purpose, including a standard feedback mechanism for quality monitoring.
Capability Tags
Capability Assessment
Purpose & Capability
The skill's name/description and included code/docs consistently implement a patent image search against an external Eureka endpoint (tool-gateway.linkfox.com). That capability aligns with the stated purpose. However, the registry metadata lists no required environment variables or primary credential while both references/api.md and the Python script expect LINKFOXAGENT_API_KEY — an incoherence between declared requirements and actual needs.
Instruction Scope
SKILL.md and the code instruct the agent to POST a JSON payload (including the image URL and search parameters) to the Eureka API. The instructions do not attempt to read unrelated local files or other credentials. Note: the skill will transmit user-supplied image URLs (and therefore the fact that the user searched that image) to an external service; this is expected for a remote search but important for privacy.
Install Mechanism
There is no install spec (instruction-only) and the included Python script is small and self-contained. No downloads from arbitrary URLs or package installs are present. This is low-risk from an installation/execution mechanism perspective.
Credentials
The code and API docs require a single environment variable LINKFOXAGENT_API_KEY used as Authorization for the tool-gateway.linkfox.com endpoint. That credential is proportionate to the service, but the registry metadata did not declare any required env vars or primary credential — an important omission. Also, the skill will send the API key as an Authorization header and the supplied image URL to external domains; if the API key has broad privileges or the images are sensitive, this has privacy/security implications.
Persistence & Privilege
The skill does not request permanent 'always' inclusion and does not modify other skills or system-wide settings. It only requires an API key (per code) to call the external API; autonomous invocation is enabled by default but is not combined here with elevated privileges.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install linkfox-eureka-patent-image-search - After installation, invoke the skill by name or use
/linkfox-eureka-patent-image-search - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release
Metadata
Frequently Asked Questions
What is Eureka Patent Image Search?
通过Eureka专利平台进行专利图像检索（以图搜图），上传一张图片URL即可检索外观设计或实用新型的相似专利。当用户提到专利图像检索、以图搜专利、外观设计搜索、图片相似专利、设计专利检索、图像搜索专利、专利视觉搜索、patent image search, search patent by image, desi... It is an AI Agent Skill for Claude Code / OpenClaw, with 61 downloads so far.
How do I install Eureka Patent Image Search?
Run "/install linkfox-eureka-patent-image-search" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Eureka Patent Image Search free?
Yes, Eureka Patent Image Search is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Eureka Patent Image Search support?
Eureka Patent Image Search is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Eureka Patent Image Search?
It is built and maintained by linkfox-ai (@linkfox-ai); the current version is v1.0.0.
More Skills