prior-art-search

Search patent databases and academic literature for prior art relevant to an invention. Use when user says "现有技术检索", "prior art search", "专利检索", "check patents", or wants to find relevant prior art.

Quality

85%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

80%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-structured, actionable search workflow with concrete query templates and a complete report schema; its main gaps are the absence of explicit validate/retry feedback loops and an inline-heavy structure that underuses progressive disclosure.

Suggestions

Add an explicit feedback loop, e.g. after Step 2/3: "If fewer than N relevant results, broaden keywords or search adjacent IPC/CPC classes and re-run."

Move the full report template (the ~35-line markdown block in Step 6) into a `references/report-template.md` and reference it, keeping SKILL.md as an overview.

Add a brief verification gate before output, e.g. "Confirm every citation has a retrievable source; mark any unsourced claim with [VERIFY] before writing the report."

Dimension	Reasoning	Score
Conciseness	The body is dense and actionable without explaining concepts Claude already knows (no "what is a patent" padding); it uses concrete query templates and a copy-paste output schema, with only minor non-essential lines such as "Adapted from `/research-lit`".	3 / 3
Actionability	It provides executable guidance throughout: concrete WebSearch/WebFetch query templates (e.g. `site:patents.google.com [keywords]`), specific extraction fields, and a fully specified markdown report template that is copy-paste ready.	3 / 3
Workflow Clarity	The six steps are clearly numbered and a Key Rules checklist plus the [VERIFY] convention provide some validation, but there is no explicit validate-and-retry feedback loop (e.g. expand keywords if a search yields nothing), keeping it at 'sequence present but checkpoints implicit'.	2 / 3
Progressive Disclosure	Section headers organize the body and the one external reference ("Load `../shared-references/prior-art-databases.md`") is clearly signaled and one level deep, but the large inline output template and search-strategy detail could be split into reference files; no bundle files are provided to offset this.	2 / 3
	Total	10 / 12 Passed

Description

90%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description with explicit what-and-when structure and natural bilingual trigger terms; its only weakness is that it names the search domain rather than enumerating the full set of concrete actions the skill performs.

Suggestions

Expand the action list to surface the full capability set, e.g. "Search patent databases and academic literature, extract claims and classifications, assess overlap risk, and generate a prior-art report."

Dimension	Reasoning	Score
Specificity	The description names two concrete domains ("patent databases and academic literature") and a single action ("Search ... for prior art"), but does not enumerate multiple specific actions like classify, assess overlap, or generate a report, so it stops at the 'names domain and some actions' anchor rather than the comprehensive list of a 3.	2 / 3
Completeness	It explicitly states what the skill does ("Search patent databases and academic literature for prior art relevant to an invention") and when to use it ("Use when user says ..."), clearly answering both what and when with explicit triggers.	3 / 3
Trigger Term Quality	It lists natural user-facing phrases in both English and Chinese ("prior art search", "check patents", "专利检索", "现有技术检索", "find relevant prior art"), giving strong coverage of terms a user would actually say.	3 / 3
Distinctiveness Conflict Risk	The patent/literature prior-art niche with bilingual, patent-specific trigger phrases is clearly distinguishable from other skills and unlikely to fire for the wrong skill.	3 / 3
	Total	11 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: wanshuiyin/Auto-claude-code-research-in-sleep
Path: skills/prior-art-search/SKILL.md
Commit: c5f3d5b

Reviewed: about 1 hour ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.