Preface¶

The common approach to prior art search (novelty search) before filing a invention patent application is step-by-step: first, ask the model for search terms, then manually search various databases, and finally compare similar patents and compile a report. Every step in the middle is prone to breakage: inaccurate search queries, unconfigured data sources, or forgetting to check for prior publications by the inventors themselves. The Skill introduced below compresses this chain into a single execution.

What is this¶

“Patent Drafting - Novelty Search” (Coordinate @user_60f0359b/patent-novelty-full-workflow-oneshot) is a Professional Skill from a community source, published by user_60f0359b, version 1.0.3. It is positioned as a “Full Patent Novelty Search Workflow (One-shot Mode)”: fully automated execution from the start of the search to the output of the report without pausing or waiting for user feedback, outputting a complete 7-chapter novelty search report in one go.

According to the official description, it targets novelty searches prior to Chinese invention patent applications. Inputs can be technical disclosures (the five documents of a patent application) and/or similar files provided by the user. The tag requires_api_key=true indicates that at least some capabilities require an API Key.

Core Features¶

According to the official description, this Skill provides the following capabilities (all below are from the Skill’s description; the pre-fetched page content did not expand on them directly, so it is recommended to verify during use):

1. Generate search keywords and search queries based on technical disclosures or similar files.
2. Perform searches for the inventors’ own prior publications, i.e., check if the applicant/inventor has published anything previously.
3. Perform comparative analysis of similar patents/papers.
4. Output a complete novelty search report consisting of 7 chapters.

Data sources are organized by “channels,” all of which are optional and configured/detected independently per channel. At runtime, check_environment.py provides a “Configured/Not Configured” checklist across four dimensions: Papers vs. Patents, Key-free vs. API-required, Current Status, and Paid vs. Free.

It defaults to “High Recall Mode”: in addition to user-configured channels, the AI will automatically attempt other key-free, free channels (OpenAlex / Crossref / Semantic Scholar / arXiv / PubMed / Europe PMC / USPTO PatentsView) with the goal of maximizing recall. It only switches to “Strict Mode” when the user explicitly states “based only on configured channels.”

Available for use offline or with poor network conditions: all dependencies are packaged into a local wheel, with no downloads required; it requires minimal file reading and provides search query suggestions (this will be marked as “Not a real search” at this time).

Trigger words include: Novelty search, help me search novelty, generate search queries, search keywords, check inventor prior publication, similar patent comparison, novelty report, prior art search.

Installation and Activation¶

The pre-fetched materials did not provide installation commands or directory conventions, so no speculation is made here. Please confirm on the official page before enabling:

https://www.skillhub.cn/skills/user_60f0359b/patent-novelty-full-workflow-oneshot

Confirmed metadata is as follows:

Name:      Patent Drafting - Novelty Search
Coordinate: @user_60f0359b/patent-novelty-full-workflow-oneshot
Version:   1.0.3
Source:    community
Category:  professional
Publisher: user_60f0359b
Tags:      requires_api_key=true

This Skill is based on the general SKILL.md format. The pre-fetched materials did not specify how to load it in Cursor / Codex CLI / Claude Code, so no fabrication is made here.

Typical Usage¶

According to the official description, a typical call works as follows:

1. Provide the technical disclosure (five documents) or similar files to the Skill.
2. Confirm data channels first. To let the AI automatically supplement key-free channels, use the default High Recall Mode; to use only configured channels, explicitly state “based only on configured channels.”
3. Upon execution, directly obtain search keywords, search queries, prior publication results, similar patent/paper comparisons, and a complete 7-chapter novelty report.
4. In an offline environment, accept the degraded output of “file reading + search query suggestions,” noting that this output will be marked as not a real search.

Applicable Scenarios and Notes¶

Suitable for: Novelty searches prior to Chinese invention patent applications, especially scenarios where the user wants to combine search, prior publication screening, similarity comparison, and report writing into a single step without stopping midway for feedback.

Points to note:

1. At least some channels require an API Key (requires_api_key=true); unconfigured channels will be displayed as unconfigured in the list provided by check_environment.py.
2. High Recall Mode calls multiple external free channels, offering maximum recall but higher time consumption and network dependency; Strict Mode is enabled only upon your explicit request.
3. Offline degraded output is merely search query suggestions, not actual search results, and cannot be directly used as a basis for novelty search.
4. The download count (1635), install count (0), and rating (100000) in the page metadata are from the fetched page (fetch time 2026-10-30) and do not constitute proof of quality.

Conclusion¶

The value of this Skill lies in merging multiple stages of novelty search into a single execution and explicitly exposing “which data sources are available, whether a Key is needed, and whether it is free or paid” via a channel list, rather than hiding them within the model’s own judgment. To verify installation methods and channel configuration details, start from the official page:

https://api.skillhub.cn/user_60f0359b/patent-novelty-full-workflow-oneshot