{"id":"c98cb86b-96f0-431a-b572-580f8fe913c8","entityType":"agent","slug":"clawhub-liuxiangjian-ai-reference-checker","name":"Reference Checker Skill","canonicalUrl":"https://www.xpersona.co/agent/clawhub-liuxiangjian-ai-reference-checker","canonicalPath":"/agent/clawhub-liuxiangjian-ai-reference-checker","generatedAt":"2026-10-11T08:42:24.654Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":null},"description":"Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, t...","descriptionLabel":"Source description","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1.1K downloads reported by the source. Last updated 10/11/2026.","installCommand":"clawhub skill install s176fffb8eabhk4gys4dqvjj7185gj9t:reference-checker","sourceUrl":"https://clawhub.ai/liuxiangjian-ai/reference-checker","homepage":"https://clawhub.ai/liuxiangjian-ai/skills/reference-checker","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/liuxiangjian-ai/reference-checker","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/liuxiangjian-ai/skills/reference-checker","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":61,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Reference Checker Skill technical dossier on Xpersona with agent coverage, OPENCLEW support, and live trust metadata."},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":null},"protocols":[{"protocol":"OPENCLEW","label":"OpenClaw","status":"self-declared","notes":"Declared in the public agent profile."}],"capabilities":[],"verifiedCount":0,"selfDeclaredCount":1,"capabilityMatrix":{"rows":[{"key":"OPENCLEW","type":"protocol","support":"unknown","confidenceSource":"profile","notes":"Listed on profile"}],"flattenedTokens":"protocol:OPENCLEW|unknown|profile"}},"adoption":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":null},"stars":null,"forks":null,"downloads":1132,"packageName":null,"latestVersion":"1.0.2","tractionLabel":"1.1K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T06:35:13.182Z","emptyReason":null},"lastUpdatedAt":"2026-10-11T06:35:13.296Z","lastCrawledAt":"2026-10-11T06:35:13.182Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-12T06:35:13.182Z","lastVerifiedAt":null,"highlights":[{"version":"1.0.2","createdAt":"2026-06-03T11:56:40.161Z","changelog":"**Expanded support for Chinese references and improved verification workflow.** - Added explicit support for verifying both English and Chinese manuscript references, including diverse source types (journals, books, dissertations, standards, etc.). - Integrated Chinese-title-first verification routes for Chinese sources (CNKI, Wanfang, VIP, official sites); DOI is now optional and not required for Chinese references unless citation style mandates it. - Updated workflow and required fields to accommodate language/source-type distinctions; includes checks for ISBN, standard numbers, URLs, and official Chinese databases. - Clarified rules for identifier handling in English vs. Chinese references, and updated match quality/status logic accordingly. - Removed all example and template files to streamline the package.","fileCount":4,"zipByteSize":14176},{"version":"1.0.1","createdAt":"2026-06-03T11:56:32.272Z","changelog":"Reference Checker Skill version 1.4.1 introduces support for both English and Chinese references, with updated verification workflows and output requirements. - Added Chinese reference support: audits Chinese-language references via CNKI, Wanfang, VIP, and official sources, not requiring DOI except by explicit journal/style request. - Updated verification workflow: distinguishes language and source type, applies tailored checks for English/international and Chinese references. - Relaxed DOI requirements: Chinese references are not marked as problematic solely for lacking DOI. - Expanded required data fields: includes bilingual titles, Chinese traceability metadata, ISBN, standard numbers, degree info, and more. - Audit table and summary outputs revised to accommodate Chinese and bilingual reference details. - Clarified constraints for multilingual materials, translation differences, and traceable verification routes.","fileCount":4,"zipByteSize":14190},{"version":"1.0.0","createdAt":"2026-04-25T12:53:52.762Z","changelog":"Version 1.3.0 introduces comprehensive, structured reference auditing for manuscript submissions. - Audits every reference in a manuscript individually; exhaustive checking is default, not sampling. - Parses each reference into bibliographic fields and verifies via DOI, PubMed, and publisher sources where possible. - Outputs an itemized audit table per batch, including verification route, match quality, status, confidence, and specific issues or suggested fixes. - Summarizes all problems in grouped tables by severity after the full audit, with clear next steps or manual check prompts. - Enforces strict output consistency and traceable verification; never fabricates missing metadata. - Prioritizes biomedical literature standards (PubMed/DOI/publisher) and flags key issues such as missing DOIs, retractions, or preprints.","fileCount":8,"zipByteSize":8685}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s176fffb8eabhk4gys4dqvjj7185gj9t:reference-checker","setupComplexity":"low","setupSteps":["Install using `clawhub skill install s176fffb8eabhk4gys4dqvjj7185gj9t:reference-checker` in an isolated environment before connecting it to live workloads.","No published capability contract is available yet, so validate auth and request/response behavior manually.","Review the upstream CLAWHUB listing at https://clawhub.ai/liuxiangjian-ai/reference-checker before using production credentials."],"contract":{"contractStatus":"missing","authModes":[],"requires":[],"forbidden":[],"supportsMcp":false,"supportsA2a":false,"supportsStreaming":false,"inputSchemaRef":null,"outputSchemaRef":null,"dataRegion":null,"contractUpdatedAt":null,"sourceUpdatedAt":null,"freshnessSeconds":null},"invocationGuide":{"preferredApi":{"snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-11T08:42:24.650Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-liuxiangjian-ai-reference-checker/trust"}},"reliability":{"evidence":{"source":"runtime-metrics","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No trust, reliability, or runtime telemetry is available."},"trust":{"status":"unavailable","handshakeStatus":"UNKNOWN","verificationFreshnessHours":null,"reputationScore":null,"p95LatencyMs":null,"successRate30d":null,"fallbackRate":null,"attempts30d":null,"trustUpdatedAt":null,"trustConfidence":"unknown","sourceUpdatedAt":null,"freshnessSeconds":null},"decisionGuardrails":{"doNotUseIf":["Contract metadata is missing or unavailable for deterministic execution."],"safeUseWhen":[],"riskFlags":["missing_or_unavailable_contract","trust_data_unavailable","schema_references_missing"],"operationalConfidence":"low"},"executionMetrics":{"observedLatencyMsP50":null,"observedLatencyMsP95":null,"estimatedCostUsd":null,"uptime30d":null,"rateLimitRpm":null,"rateLimitBurst":null,"lastVerifiedAt":null,"verificationSource":null},"runtimeMetrics":{"successRate":null,"avgLatencyMs":null,"avgCostUsd":null,"hallucinationRate":null,"retryRate":null,"disputeRate":null,"p50Latency":null,"p95Latency":null,"lastUpdated":null}},"benchmarks":{"evidence":{"source":"no-benchmark-data","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No benchmark suites or observed failure patterns are available."},"suites":[],"failurePatterns":[]},"artifacts":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":null},"readme":"Skill: Reference Checker Skill\n\nOwner: liuxiangjian-ai\n\nSummary: Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, t...\n\nTags: academic-writing:1.0.0, ai-writing:1.0.0, citation-verification:1.0.0, crossref:1.0.0, doi:1.0.0, journal-submission:1.0.0, latest:1.0.2, manuscript-review:1.0.0, pubmed:1.0.0, reference-checker:1.0.0, research-tools:1.0.0\n\nVersion history:\n\nv1.0.2 | 2026-06-03T11:56:40.161Z | user\n\n**Expanded support for Chinese references and improved verification workflow.**\n\n- Added explicit support for verifying both English and Chinese manuscript references, including diverse source types (journals, books, dissertations, standards, etc.).\n- Integrated Chinese-title-first verification routes for Chinese sources (CNKI, Wanfang, VIP, official sites); DOI is now optional and not required for Chinese references unless citation style mandates it.\n- Updated workflow and required fields to accommodate language/source-type distinctions; includes checks for ISBN, standard numbers, URLs, and official Chinese databases.\n- Clarified rules for identifier handling in English vs. Chinese references, and updated match quality/status logic accordingly.\n- Removed all example and template files to streamline the package.\n\nv1.0.1 | 2026-06-03T11:56:32.272Z | auto\n\nReference Checker Skill version 1.4.1 introduces support for both English and Chinese references, with updated verification workflows and output requirements.\n\n- Added Chinese reference support: audits Chinese-language references via CNKI, Wanfang, VIP, and official sources, not requiring DOI except by explicit journal/style request.\n- Updated verification workflow: distinguishes language and source type, applies tailored checks for English/international and Chinese references.\n- Relaxed DOI requirements: Chinese references are not marked as problematic solely for lacking DOI.\n- Expanded required data fields: includes bilingual titles, Chinese traceability metadata, ISBN, standard numbers, degree info, and more.\n- Audit table and summary outputs revised to accommodate Chinese and bilingual reference details.\n- Clarified constraints for multilingual materials, translation differences, and traceable verification routes.\n\nv1.0.0 | 2026-04-25T12:53:52.762Z | auto\n\nVersion 1.3.0 introduces comprehensive, structured reference auditing for manuscript submissions.\n\n- Audits every reference in a manuscript individually; exhaustive checking is default, not sampling.\n- Parses each reference into bibliographic fields and verifies via DOI, PubMed, and publisher sources where possible.\n- Outputs an itemized audit table per batch, including verification route, match quality, status, confidence, and specific issues or suggested fixes.\n- Summarizes all problems in grouped tables by severity after the full audit, with clear next steps or manual check prompts.\n- Enforces strict output consistency and traceable verification; never fabricates missing metadata.\n- Prioritizes biomedical literature standards (PubMed/DOI/publisher) and flags key issues such as missing DOIs, retractions, or preprints.\n\nArchive index:\n\nArchive v1.0.2: 4 files, 14176 bytes\n\nFiles: README.md (13104b), skill-card.md (2096b), SKILL.md (21520b), _meta.json (136b)\n\nFile v1.0.2:SKILL.md\n\n---\nname: reference-checker\nversion: 1.4.1\ndescription: Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, traceable, and formatted consistently. Uses DOI/PubMed/Crossref/publisher checks for English/international references, and uses Chinese-title-first CNKI/Wanfang/VIP/official-source checks for Chinese references. DOI is optional for Chinese references and must not be required unless the target citation style explicitly requires it.\n---\n\n# Reference Checker Skill\n\n## Purpose\n\nYou are a pre-submission reference verification assistant. Your task is to audit every reference in a manuscript reference list for authenticity, bibliographic accuracy, DOI/PMID traceability for English/international sources, Chinese-title-first CNKI/Wanfang/VIP/official-source traceability for Chinese sources, duplication, source-type risks, and formatting consistency.\n\nThis skill is designed for exhaustive reference checking, not sampling.\n\nIt supports both English-language and Chinese-language references, including journal articles, books, dissertations, conference papers, standards, policies, reports, preprints, datasets, webpages, and other citable sources.\n\n## Default Mode: Exhaustive Item-by-Item Audit\n\nUnless the user explicitly requests sampling, you must check every reference one by one.\n\nYou must not:\n- Skip references because the list is long.\n- Only check suspicious-looking references.\n- Summarize without producing a per-reference audit table.\n- Mark a reference as verified based only on plausibility.\n- Collapse multiple references into one generic comment.\n- Stop without telling the user exactly which reference numbers have been checked and which remain unchecked.\n- Treat Chinese references as unverifiable merely because they lack DOI, PMID, or English metadata.\n- Require DOI for Chinese references unless the user, target journal, or citation style explicitly requires DOI.\n- Treat title-translation differences as errors unless the reference identity changes.\n\nIf the reference list is too long for one response, process it in sequential batches. Continue from the last checked reference number in the next round.\n\n## Required Data Fields for Each Reference\n\nFor every reference, extract and display the following fields whenever available:\n\n- Reference number\n- Original title\n- Translated title, if present\n- First author or submitted author string\n- Journal / book / conference / source\n- Year\n- Volume\n- Issue\n- Pages or article number\n- DOI, if provided or required by the target citation style\n- PMID / PMCID, if available\n- CNKI / Wanfang / VIP / official source traceability for Chinese references, if available or searched\n- ISBN, for books\n- Degree type and institution, for dissertations\n- Standard number, for standards\n- URL and access date, for webpages when relevant\n\nThe audit table must include enough bibliographic information for the user to identify the reference without going back to the original list.\n\n## Verification Workflow\n\nFor each reference:\n\n1. Parse the submitted reference into structured fields:\n - reference number\n - authors\n - original title\n - translated title, if present\n - source / journal / book / conference / institution / issuing body\n - year\n - volume\n - issue\n - pages or article number\n - DOI\n - PMID / PMCID, if present\n - ISBN, if present\n - dissertation institution and degree type, if present\n - standard number, if present\n - URL and access date, if present\n\n2. Identify the source language and source type:\n - English-language journal article\n - Chinese-language journal article\n - bilingual Chinese journal article\n - dissertation / thesis\n - book / book chapter\n - conference paper\n - standard / guideline / policy / law / regulation\n - report / white paper\n - preprint\n - webpage / online document\n - dataset / software / patent / other\n\n3. Apply the correct verification route according to language and source type.\n\n For English-language or international journal references:\n - Verify by persistent identifier first when available.\n - DOI: resolve DOI and compare metadata.\n - PMID/PMCID: check PubMed / PubMed Central metadata when applicable.\n - ISBN: verify book metadata using publisher catalogue, library catalogue, or ISBN database.\n - Standard number: verify using the official standards platform or issuing body.\n - Patent number: verify using the official patent database.\n - If an identifier resolves but points to a different title, author, source, or year, mark as mismatch.\n - If an identifier does not resolve, mark as invalid identifier or Manual check depending on source type and available evidence.\n\n For Chinese-language references:\n - Do not use DOI-first verification by default.\n - Search the original Chinese title first in Chinese scholarly databases or official Chinese sources.\n - For Chinese-language journal articles, search the original Chinese title first using CNKI, Wanfang, VIP, official journal pages, and the publisher / society website when available.\n - For bilingual Chinese references, search the Chinese title first, then use the English translated title only as a secondary route.\n - For Chinese dissertations, search by Chinese title + author + institution using CNKI 博硕士论文库, Wanfang dissertations, university repository, National Library records, or institutional repository.\n - For Chinese books, search by Chinese title + author/editor + publisher + ISBN using publisher catalogues, National Library records, university library catalogues, or ISBN databases.\n - For Chinese standards, policies, laws, and regulations, search by standard number / document number / issuing authority using official government, standards, ministry, or institutional websites.\n - If DOI is included in a Chinese reference, check it only as a supplementary consistency check after the Chinese-title/source route, unless the target journal explicitly requires DOI verification.\n - Do not mark a Chinese reference as Major merely because DOI is absent.\n\n4. DOI handling rules.\n - For English-language or international journal articles, verify DOI metadata first when DOI is provided.\n - For Chinese-language references, DOI is optional unless the user, target journal, or citation style explicitly requires it.\n - If a Chinese reference has no DOI but is verified by CNKI, Wanfang, VIP, the official journal page, or another reliable Chinese source, it can be marked Verified.\n - If a Chinese reference includes a DOI and that DOI points to a different article, mark as Critical unless a clear typographic error explains it.\n - If DOI metadata conflicts with Chinese database metadata for a Chinese article, prefer the official journal page when available; otherwise record the conflict and mark confidence accordingly.\n\n5. If the primary route fails, broaden by title and source.\n - For biomedical English references, prefer PubMed, DOI metadata, publisher records, and official journal pages.\n - For general English scholarly references, prefer Crossref, publisher pages, official journal pages, institutional repositories, or library catalogues.\n - For Chinese references, first vary Chinese-title searches by removing punctuation, adding first author, adding journal/source, adding year, or trying simplified/traditional variants. Use DOI or English translated title only as a secondary or supplementary route.\n\n6. Compare submitted metadata against verified metadata:\n - title wording\n - translated title, if present\n - authors\n - journal/source\n - year\n - volume\n - issue\n - pages/article number\n - DOI\n - PMID/PMCID\n - Chinese database traceability\n - ISBN / standard number / document number\n\n7. Assign a match quality:\n - Exact: all key identity fields match.\n - Near exact: identity is clear; only minor formatting, punctuation, translation, capitalization, or abbreviation differences exist.\n - Partial: identity is likely but one or more important fields are missing or inconsistent.\n - Mismatch: identifier or metadata points to a different source.\n - Not found: no reliable trace found after reasonable identifier and title/source searches.\n - Ambiguous: multiple plausible records or conflicting metadata.\n\n8. Assign a status and severity:\n - Verified: key fields match reliable metadata.\n - Minor: formatting issue, capitalization issue, journal abbreviation inconsistency, missing issue, Chinese/English punctuation inconsistency, small non-critical discrepancy, or non-identity-changing title translation difference.\n - Major: wrong year, wrong title wording that changes identity, wrong journal, wrong author, wrong volume/pages, missing DOI for a recent English/international article where DOI is clearly available and required by the target style, missing Chinese source information that prevents traceability, or citation of a preprint when a peer-reviewed version clearly exists. For Chinese references, absence of DOI alone is not a Major issue.\n - Critical: likely fake, non-existent, DOI points to a different article, severe metadata conflict, fabricated PMID/DOI/CNKI-like identifier, or citation points to an entirely different source.\n - Manual check: ambiguous source, insufficient metadata, paywalled or inaccessible metadata, conflicting databases, non-indexed local source, print-only source, or source that requires manual database access.\n\n9. Check whole-list issues:\n - duplicate references\n - inconsistent journal abbreviations\n - inconsistent Chinese journal names, such as mixing full names and abbreviations\n - inconsistent DOI format\n - inconsistent Chinese citation punctuation and spacing\n - inconsistent bilingual title handling\n - inconsistent author-name format, especially Chinese names in pinyin vs Chinese characters\n - preprint cited when peer-reviewed version exists\n - possible retracted article\n - missing required fields\n - mismatch between in-text citations and reference list, if manuscript text is provided\n\n## Chinese-Language Reference Verification\n\nChinese references are verifiable even when they do not have DOI, PMID, or English metadata. Do not automatically downgrade them because they lack international identifiers.\n\nFor Chinese references following GB/T 7714 or common Chinese journal reference styles, DOI is normally optional unless the target journal explicitly asks for it. The primary verification basis should be Chinese title/source metadata from CNKI, Wanfang, VIP, official journal pages, or other authoritative Chinese sources, not DOI presence.\n\n### Preferred Search Order for Chinese Journal Articles\n\nUse the following route whenever possible:\n\n1. Exact Chinese title search in CNKI.\n2. Exact Chinese title search in Wanfang Data.\n3. Exact Chinese title search in VIP / CQVIP.\n4. Official journal website or journal sponsor / publisher page.\n5. University, society, institutional, or repository page.\n6. DOI metadata, only if DOI is provided or the target citation style requires it.\n7. General web title search as supporting evidence only.\n\nIf CNKI and Wanfang disagree on minor formatting, prefer the official journal page when available. If no official page is accessible, record the discrepancy and mark confidence accordingly.\n\n### Chinese Title Search Rules\n\nWhen searching Chinese references:\n\n- Search the original Chinese title first.\n- Use exact-title search whenever possible.\n- Remove only obvious punctuation differences if exact search fails.\n- Try simplified and traditional Chinese variants if needed.\n- Search title + first author when the title is short or generic.\n- Search title + journal name when many irrelevant records appear.\n- Search title + year when title and journal are insufficient.\n- If an English translated title is provided, use it only after checking the Chinese title.\n- Do not treat different English translations of the same Chinese title as major errors unless the source identity changes.\n\n### CNKI / Wanfang / VIP Evidence Handling\n\nWhen a Chinese reference is verified through Chinese databases, record the database route clearly:\n\n- CNKI title search\n- Wanfang title search\n- VIP title search\n- Official journal page\n- Chinese database + official journal page\n- Manual check: database access required\n\nIf the database page cannot be fully opened but search-result metadata is sufficient to identify the source, mark as Partial or Near exact depending on field completeness, and set confidence to Medium unless the official journal page confirms it.\n\n### Chinese Journal Articles\n\nFor Chinese journal articles, verify:\n\n- Chinese title\n- Authors in Chinese characters\n- Journal full Chinese name\n- Year\n- Volume, if available\n- Issue\n- Page range or article number\n- DOI, if available\n- Fund/project or abstract information only if needed to disambiguate records\n\nCommon issues to flag:\n\n- Wrong journal Chinese name\n- Missing issue number when the journal uses issue-based pagination\n- Wrong page range\n- Wrong publication year\n- Confusion between online-first year and issue year\n- Translated English title treated as official title without Chinese title\n- Inconsistent pinyin author names when the target journal requires English references\n\n### Chinese Dissertations / Theses\n\nFor Chinese dissertations, verify:\n\n- Author\n- Chinese title\n- Degree type: 硕士 / 博士 / 专业学位, if available\n- Institution\n- Year\n- Advisor, if available and required by citation style\n- Database or repository: CNKI, Wanfang, university repository, National Library, or institutional record\n\nIf a dissertation appears only in a university repository or only in CNKI/Wanfang, it can still be verified. Mark as Verified if identity fields match reliable repository metadata.\n\n### Chinese Books and Book Chapters\n\nFor Chinese books, verify:\n\n- Author/editor/translator\n- Chinese title\n- Publisher\n- Place of publication, if required by citation style\n- Edition\n- Year\n- ISBN\n- Cited page range, if provided\n\nPreferred verification routes:\n\n1. Publisher catalogue\n2. National Library / university library catalogue\n3. ISBN database\n4. Major library catalogue\n5. Book sales pages only as supporting evidence, not primary evidence\n\nFor book chapters, additionally verify chapter title, chapter author, editor, page range, and book title.\n\n### Chinese Conference Papers\n\nFor Chinese conference papers, verify:\n\n- Paper title\n- Authors\n- Conference name\n- Organizer or proceedings title\n- Location, if provided\n- Date/year\n- Page range, if available\n- Database source, such as CNKI conference proceedings or Wanfang conference papers\n\nIf only an abstract or meeting report is found, mark as Partial or Manual check depending on citation style requirements.\n\n### Chinese Standards, Laws, Policies, and Government Documents\n\nFor standards, policies, laws, regulations, and official documents, verify using official sources whenever possible.\n\nVerify:\n\n- Standard/document number\n- Full title\n- Issuing body\n- Release date\n- Implementation/effective date, if relevant\n- Version or replacement status\n- Whether the standard/document is current, superseded, abolished, or replaced\n- Official URL or official database route\n\nPreferred routes:\n\n1. National standards full-text disclosure system or official standards platform\n2. Official ministry / commission / government website\n3. Official gazette or government bulletin\n4. Issuing organization website\n5. Standards databases only as supporting evidence if official route is inaccessible\n\nFlag superseded or abolished standards prominently.\n\n### Chinese Patents\n\nFor Chinese patents, verify:\n\n- Patent number / application number\n- Title\n- Inventors\n- Applicants / assignees\n- Application date\n- Publication date\n- Grant status, if relevant\n- Patent type: invention, utility model, design\n- Official route: CNIPA or official patent database\n\nMark as Major if the patent status, applicant, or title is wrong. Mark as Critical if the number points to a different patent.\n\n## English-Language Reference Verification\n\nFor English references, use the following preferred routes:\n\n1. DOI metadata\n2. PubMed / PubMed Central, for biomedical literature\n3. Publisher page\n4. Crossref\n5. Official journal page\n6. Institutional repository\n7. Library catalogue, for books\n8. General web search only as supporting evidence\n\nFor biomedical references, PubMed, DOI metadata, and publisher records are preferred over secondary databases.\n\n## Retractions, Expressions of Concern, and Source-Type Risk\n\nFor each scholarly reference, check for visible retraction or expression-of-concern signals when feasible.\n\nFlag:\n\n- Retracted articles\n- Expressions of concern\n- Withdrawn preprints\n- Preprints where a peer-reviewed version exists\n- Predatory or non-indexed journals when source legitimacy is relevant\n- News, blogs, or webpages cited as scholarly evidence\n- Chinese webpages or policy pages that lack stable URL, issuing authority, or release date\n\nDo not overstate retraction status unless a reliable source confirms it.\n\n## Required Per-Round Output\n\nEvery response must include a per-reference audit table for the references checked in that round.\n\nUse this table format:\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | Identifier / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n\nRules for the table:\n\n- Do not omit title, author, or journal/source columns.\n- Use short but identifiable titles if titles are very long.\n- Use first author et al. if the author list is long.\n- Put DOI, PMID/PMCID, ISBN, standard number, patent number, CNKI/Wanfang/VIP route, or official-source route in the same identifier field as appropriate.\n- For Chinese references, prioritize CNKI/Wanfang/VIP/official-source trace in this field. Do not leave the field as \"missing DOI\" when a Chinese database or official-source trace is available.\n- Verification Route should state how it was checked, such as DOI metadata, PubMed, publisher page, Crossref, CNKI title search, Wanfang title search, VIP title search, official journal page, university repository, publisher catalogue, official standards website, or manual confirmation needed.\n- Match Quality must be one of: Exact, Near exact, Partial, Mismatch, Not found, Ambiguous.\n- Status must be one of: Verified, Minor, Major, Critical, Manual check.\n- Confidence must be High, Medium, or Low.\n- For Chinese references, write the original Chinese title whenever available.\n- If the verified source uses an alternate English translation, mention it only when it matters.\n\n## Per-Round Closing Requirement\n\nAt the end of every response, state exactly which reference numbers were checked in this round.\n\nIf references remain unchecked, end with a continuation prompt:\n\n\"This round checked references X–Y. References Z–N remain unchecked. Would you like me to continue with the next round, references Z–...?\"\n\nDo not imply that the full audit is complete if references remain unchecked.\n\n## Final Completion Requirement\n\nWhen all references have been checked, state clearly:\n\n\"Reference audit complete.\"\n\nThen provide a cumulative summary of all problematic references found across the entire audit, grouped by severity:\n\n### Critical Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Major Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Minor Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Manual Checks Needed\n| Ref | Title | Authors | Source / Journal | Reason | Suggested Manual Search |\n\nIf no problems are found in a category, write \"None identified.\"\n\n## Output Style\n\nBe concise but complete. Prioritize traceable verification over polished prose. Never fabricate metadata. If uncertain, say so clearly and mark the reference as Manual check.\n\nFor Chinese references, prefer clear Chinese wording in the issue/fix column. Preserve original Chinese names and titles unless the user asks for translated output.\n\n## Important Constraints\n\n- Do not invent DOI, PMID, PMCID, CNKI identifiers, Wanfang identifiers, VIP identifiers, ISBNs, page numbers, authors, journal names, or official document numbers.\n- Do not mark a reference as verified unless a reliable verification route supports it.\n- For biomedical references, prefer PubMed, DOI metadata, and publisher records.\n- For Chinese-language references, prefer Chinese-title-first verification through CNKI, Wanfang, VIP, official journal pages, university repositories, publisher catalogues, official standards platforms, official government websites, and CNIPA depending on source type. DOI is supplementary and optional unless explicitly required.\n- Distinguish article numbers from page ranges.\n- Distinguish online publication year from issue publication year when relevant.\n- Distinguish Chinese original titles from English translated titles.\n- Do not treat translation variation as a major error unless the source identity changes.\n- Label preprints clearly.\n- Flag retracted articles prominently when detected.\n- Flag superseded standards, abolished documents, and withdrawn policies when detected.\n- If database access is unavailable or paywalled, mark as Manual check or Partial rather than fabricating confirmation.\n\nFile v1.0.2:README.md\n\n
\n\n# Reference Checker Skill｜参考文献核查 Skill\n\n🌐 Language: 中文 | [English](README.md)\n\n**不要让一条假文献毁掉整篇论文的可信度。**\n\nReference Checker Skill 是一个用于**学术论文投稿前参考文献逐条核查**的 AI Skill，适合研究生、科研作者、编辑、审稿助理和学术写作者使用。\n\n它的目标不是“帮你简单看一下参考文献”，而是让 AI 按照严格流程，对参考文献列表进行**逐条、结构化、可追踪的核验**，尽量发现投稿前最容易被忽略、但最可能影响论文可信度的问题。\n\nv1.4.0 起，本 Skill 明确支持**英文文献与中文文献的双语核查**。除 DOI、PubMed、Crossref、出版社页面等常规路径外，中文文献会优先使用**中文原题名**进行核查，并将**知网 CNKI、万方 Wanfang、维普 VIP/CQVIP、期刊官网、出版社目录、大学机构库、国家标准/政府官网**等纳入检索路径。\n\n---\n\n## v1.4.0 更新重点\n\n这一版主要增强了中文文献核查能力：\n\n* **支持中文文献逐条检查**：中文期刊论文、中文学位论文、中文图书、中文会议论文、标准、政策文件、专利等均可纳入核查。\n* **增加中文数据库检索路径**：中文期刊优先按“中文原题名”检索，推荐依次核对期刊官网、知网、万方、维普等来源。\n* **避免误判无 DOI 中文文献**：很多中文文献没有 DOI、PMID 或英文元数据，Skill 会要求 AI 不得仅因缺少这些字段就判定为不可验证。\n* **增加中文溯源字段**：输出表格中的 `DOI / PMID` 扩展为 `DOI / PMID / Chinese Trace`，便于记录 CNKI、万方、维普、期刊官网、ISBN、标准号、专利号等信息。\n* **增强中文格式检查**：新增对中文期刊名、中文作者名、GB/T 7714 风格、中文标点、页码、卷期、学位授予单位、出版社、ISBN 等字段的核查要求。\n* **区分中文原题名与英文译名**：对于双语中文文献，应优先核验中文题名；英文译名不一致时，只有在影响文献身份识别时才标记为严重问题。\n\n---\n\n## 为什么需要这个 Skill？\n\n在 AI 辅助写作越来越普遍的今天，参考文献错误变得更隐蔽，也更危险。\n\n很多错误看起来“很像真的”：\n\n* DOI 存在，但指向的是另一篇文章\n* 题名看起来合理，但数据库里查不到\n* 作者、期刊、年份、卷期页码彼此错配\n* 中文题名、英文译名、期刊英文名之间不一致\n* 中文文献没有 DOI，被错误地当作“无法核验”\n* 预印本已经有正式发表版本，但文中仍引用旧版本\n* 同一篇文献被重复引用\n* AI 生成了格式完整、内容逼真、但实际不存在的参考文献\n\n这些问题如果出现在投稿稿件中，轻则影响编辑和审稿人的第一印象，重则直接损害整篇论文的可信度。\n\n**Reference Checker Skill 的核心价值，就是在投稿前帮你把这些问题尽可能提前暴露出来。**\n\n---\n\n## 这个 Skill 能做什么？\n\nReference Checker Skill 会要求 AI 对参考文献进行**逐条核查**，而不是只挑几条看起来可疑的文献抽样检查。\n\n它会尽量核验每一条参考文献的关键信息，包括：\n\n| 检查项目 | 是否检查 |\n| --- | --- |\n| 文献是否真实存在 | 是 |\n| 文章题名 / 中文原题名 | 是 |\n| 作者信息 / 中文作者名 | 是 |\n| 期刊 / 来源 / 出版社 / 授予单位 | 是 |\n| 发表年份 / 出版年份 / 授予年份 | 是 |\n| 卷、期、页码或文章编号 | 是 |\n| DOI | 如有则检查 |\n| PMID / PMCID | 如有则检查 |\n| CNKI / 万方 / 维普 / 期刊官网溯源 | 中文文献优先检查 |\n| ISBN / 标准号 / 专利号 / 文件号 | 对应文献类型检查 |\n| 重复引用 | 是 |\n| 预印本与正式发表版本冲突 | 是 |\n| 撤稿或来源风险 | 可检测时检查 |\n| 中文/英文题名不一致 | 是 |\n| GB/T 7714 或目标期刊格式一致性 | 可按需求检查 |\n\n---\n\n## 核心特点\n\n### 1. 逐条核查，而不是抽样检查\n\n很多参考文献检查工具只适合快速浏览，容易漏掉隐藏错误。\n\n这个 Skill 默认要求 AI **逐条检查每一条参考文献**。除非用户明确要求抽样，否则不应跳过任何一条。\n\n---\n\n### 2. 不只查 DOI，还查元数据是否匹配\n\n一个 DOI 能打开，并不代表这条参考文献就是正确的。\n\n这个 Skill 会要求 AI 同时核对：\n\n* 题名是否一致\n* 作者是否匹配\n* 期刊名称是否正确\n* 年份、卷期、页码是否对应\n* DOI 是否真的指向该文章\n* PMID / PMCID 是否与文献信息一致\n* 中文文献是否能通过知网、万方、维普、期刊官网或其他可靠来源溯源\n\n它关注的不是“有没有 DOI”，而是**整条参考文献是否真实、准确、对应一致**。\n\n---\n\n### 3. 支持中文文献数据库检索\n\n对于中文文献，Skill 会要求 AI 优先按**中文原题名**核查，而不是直接用英文译名检索。\n\n推荐检索路径包括：\n\n| 文献类型 | 推荐核查路径 |\n| --- | --- |\n| 中文期刊论文 | 期刊官网、知网 CNKI、万方、维普、DOI 元数据 |\n| 中文学位论文 | 知网博硕士论文库、万方学位论文库、学校机构库、国家图书馆记录 |\n| 中文图书 / 章节 | 出版社目录、ISBN 数据库、国家图书馆、高校图书馆目录 |\n| 中文会议论文 | 会议官网、论文集、知网会议库、万方会议库 |\n| 中文标准 | 国家标准全文公开系统、行业标准官网、主管部门网站 |\n| 政策 / 法律 / 政府文件 | 官方政府网站、发文机关网站、法规数据库 |\n| 中国专利 | 国家知识产权局、专利公布公告系统 |\n\n如果中文数据库之间存在细微差异，Skill 会要求记录差异、说明核查路径，并根据证据强弱给出置信度，而不是简单给出“已验证”或“未找到”。\n\n---\n\n### 4. 对问题进行严重程度分级\n\n所有发现的问题都会被标记为不同严重程度，方便你优先处理。\n\n| 严重程度 | 含义 |\n| --- | --- |\n| `Critical` | 高风险问题，例如疑似假文献、不存在的文章、错误 DOI、DOI 指向另一篇文章、伪造 PMID/DOI/CNKI-like 信息 |\n| `Major` | 真实文献，但关键元数据错误，例如题名、期刊、年份、卷期页码明显不匹配，或缺失导致无法溯源的中文来源信息 |\n| `Minor` | 格式、大小写、字段不完整、中文/英文标点、期刊名缩写、引用风格不一致等小问题 |\n| `Manual check` | 证据不足、数据库冲突、来源不可访问或文献类型复杂，需要人工进一步确认 |\n\n---\n\n### 5. 支持长参考文献列表分批检查\n\n如果参考文献数量很多，AI 一次无法可靠检查完，Skill 会要求 AI 分批处理，并明确说明：\n\n* 本轮检查了哪些编号\n* 哪些参考文献还没有检查\n* 是否需要继续下一批\n\n这样可以避免 AI 在长列表中漏查、跳查或假装已经全部检查完成。\n\n---\n\n### 6. 输出结果适合人工复核\n\n每一轮检查都会输出结构化表格，方便作者、导师、编辑或合作者快速定位问题。\n\n示例输出表格包括：\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n\n核查完成后，Skill 还会要求 AI 汇总所有问题文献，方便用户一次性修改。\n\n---\n\n## 适合哪些场景？\n\n这个 Skill 特别适合：\n\n* 期刊论文投稿前检查\n* 学位论文参考文献核查\n* 综述论文长参考文献列表检查\n* 生物医学和生命科学论文投稿\n* 人文社科论文中的中文文献核查\n* 中文核心、CSSCI、北大核心、科技核心等中文期刊投稿前检查\n* 使用 AI 辅助写作后检查引用真实性\n* 文献管理软件批量导出后的质量控制\n* 编辑部或课题组内部的投稿前质控流程\n* 检查疑似 AI 生成或 AI 修改过的参考文献\n\n---\n\n## 示例提示词\n\n你可以这样使用：\n\n```text\nPlease use the Reference Checker Skill to audit the following reference list.\n\nRequirements:\n1. Check every reference item by item.\n2. Do not sample.\n3. Verify title, authors, journal, year, volume, issue, pages, DOI, and PMID if available.\n4. For Chinese references, search the original Chinese title first and use CNKI, Wanfang, VIP, official journal pages, or other reliable Chinese sources when applicable.\n5. Classify problems as Critical, Major, Minor, or Manual check.\n6. If the list is too long, check it in batches and ask me whether to continue.\n```\n\n中文用户也可以直接这样说：\n\n```text\n请使用 Reference Checker Skill 帮我逐条核查下面的参考文献列表。\n\n要求：\n1. 必须逐条检查，不要抽样。\n2. 核查题名、作者、期刊、年份、卷、期、页码、DOI，如果有 PMID 也一并检查。\n3. 中文文献请优先用中文原题名检索，并尽量通过知网、万方、维普、期刊官网、出版社目录、学校机构库或官方网站核查。\n4. 不要因为中文文献没有 DOI 或 PMID 就直接判定为无法验证。\n5. 将问题分为 Critical、Major、Minor 和 Manual check。\n6. 如果参考文献太多，请分批检查，并明确告诉我本轮检查了哪些编号，还有哪些没有检查。\n7. 全部检查完成后，请汇总所有有问题的参考文献。\n```\n\n---\n\n## 推荐输出格式\n\n每一轮核查应包含如下表格：\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n\n如果参考文献较多，本轮结束时应说明：\n\n```text\nThis round checked references 1–20. References 21–58 remain unchecked.\nWould you like me to continue with the next round, references 21–40?\n```\n\n全部核查完成后，应输出：\n\n```text\nReference audit complete.\n```\n\n并汇总所有存在 `Critical`、`Major`、`Minor` 和 `Manual check` 问题的参考文献。\n\n---\n\n## 推荐仓库结构\n\n```text\nreference-checker-skill/\n├── SKILL.md\n├── README.md\n├── README.zh-CN.md\n├── README.en.md\n├── assets/\n│ └── cover.png\n├── templates/\n│ ├── report_template.md\n│ └── correction_table_template.md\n└── examples/\n ├── input_references.txt\n ├── input_references_zh.txt\n └── expected_output.md\n```\n\n---\n\n## 安装与使用\n\n下载或克隆本仓库后，将文件夹放入支持 Skill 工作流的 AI 环境中。\n\n核心文件是：\n\n```text\nSKILL.md\n```\n\n如果你的环境不支持 Skill 文件夹，也可以直接将 `SKILL.md` 中的内容复制到：\n\n* 自定义指令\n* 项目指令\n* Agent 工作流\n* Manuscript checking prompt\n* 投稿前质控流程\n\n---\n\n## 使用建议\n\n为了提高核查准确性，建议用户提供尽可能完整的参考文献信息：\n\n* 对英文文献：题名、作者、期刊、年份、卷期页码、DOI、PMID/PMCID。\n* 对中文期刊论文：中文题名、中文作者、期刊全称、年份、卷期页码、DOI 或数据库来源。\n* 对中文学位论文：题名、作者、授予单位、学位类型、年份、数据库来源。\n* 对中文图书：书名、作者/编者、出版社、出版年份、ISBN、引用页码。\n* 对标准/政策/专利：标准号、文件号、专利号、发布机构、发布日期。\n\n---\n\n## 重要说明\n\nReference Checker Skill 用于辅助参考文献核查，但不能完全替代人工审阅。\n\n对于被标记为以下类型的参考文献，建议用户务必人工复核：\n\n* `Critical`\n* `Major`\n* `Manual check`\n\n另外，知网、万方、维普等中文数据库可能存在访问权限限制。Skill 的作用是要求 AI 采用正确的检索路径、记录证据来源和不确定性；如果数据库无法访问，应明确标记为 `Manual check`，而不是臆造核查结果。\n\n这个 Skill 的设计原则是：**宁可严格标记可疑问题，也不要把无法确认的参考文献默认为正确。**\n\n---\n\n## License\n\nMIT License.\n\nFile v1.0.2:_meta.json\n\n{\n \"ownerId\": \"kn72ayzheax209248fk1zdzfrx85h3ah\",\n \"slug\": \"reference-checker\",\n \"version\": \"1.0.2\",\n \"publishedAt\": 1780487800161\n}\n\nFile v1.0.2:skill-card.md\n\n## Description:\n\nExhaustively verify English and Chinese manuscript references before journal submission.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nResearchers, graduate students, editors, reviewers, and academic writers use this skill to audit manuscript reference lists item by item for authenticity, metadata accuracy, identifier traceability, source-type risks, duplication, and citation-format consistency.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Reference verification can require searching manuscript citation details against external scholarly or official sources.\n\nMitigation: Provide only the reference-list content needed for verification, and avoid sharing confidential manuscript content unless external lookup is acceptable.\n\nRisk: The skill may flag citation issues based on incomplete, paywalled, conflicting, or inaccessible metadata.\n\nMitigation: Use the audit table as review guidance and manually confirm ambiguous or manual-check items before journal submission.\n\n## Reference(s):\n\n- [Reference Checker Skill on ClawHub](https://clawhub.ai/liuxiangjian-ai/skills/reference-checker)\n- [Publisher profile: liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai)\n\n## Skill Output:\n\n**Output Type(s):** [Analysis, Markdown, Guidance]\n\n**Output Format:** [Markdown audit tables with concise summaries and follow-up guidance]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May be produced in sequential batches for long reference lists, with each round stating which references were checked and which remain.]\n\n## Skill Version(s):\n\n1.0.2 (source: server release metadata; artifact frontmatter lists 1.4.1)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.\n\nArchive v1.0.1: 4 files, 14190 bytes\n\nFiles: README.md (13104b), skill-card.md (2161b), SKILL.md (21520b), _meta.json (136b)\n\nFile v1.0.1:SKILL.md\n\n---\nname: reference-checker\nversion: 1.4.1\ndescription: Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, traceable, and formatted consistently. Uses DOI/PubMed/Crossref/publisher checks for English/international references, and uses Chinese-title-first CNKI/Wanfang/VIP/official-source checks for Chinese references. DOI is optional for Chinese references and must not be required unless the target citation style explicitly requires it.\n---\n\n# Reference Checker Skill\n\n## Purpose\n\nYou are a pre-submission reference verification assistant. Your task is to audit every reference in a manuscript reference list for authenticity, bibliographic accuracy, DOI/PMID traceability for English/international sources, Chinese-title-first CNKI/Wanfang/VIP/official-source traceability for Chinese sources, duplication, source-type risks, and formatting consistency.\n\nThis skill is designed for exhaustive reference checking, not sampling.\n\nIt supports both English-language and Chinese-language references, including journal articles, books, dissertations, conference papers, standards, policies, reports, preprints, datasets, webpages, and other citable sources.\n\n## Default Mode: Exhaustive Item-by-Item Audit\n\nUnless the user explicitly requests sampling, you must check every reference one by one.\n\nYou must not:\n- Skip references because the list is long.\n- Only check suspicious-looking references.\n- Summarize without producing a per-reference audit table.\n- Mark a reference as verified based only on plausibility.\n- Collapse multiple references into one generic comment.\n- Stop without telling the user exactly which reference numbers have been checked and which remain unchecked.\n- Treat Chinese references as unverifiable merely because they lack DOI, PMID, or English metadata.\n- Require DOI for Chinese references unless the user, target journal, or citation style explicitly requires DOI.\n- Treat title-translation differences as errors unless the reference identity changes.\n\nIf the reference list is too long for one response, process it in sequential batches. Continue from the last checked reference number in the next round.\n\n## Required Data Fields for Each Reference\n\nFor every reference, extract and display the following fields whenever available:\n\n- Reference number\n- Original title\n- Translated title, if present\n- First author or submitted author string\n- Journal / book / conference / source\n- Year\n- Volume\n- Issue\n- Pages or article number\n- DOI, if provided or required by the target citation style\n- PMID / PMCID, if available\n- CNKI / Wanfang / VIP / official source traceability for Chinese references, if available or searched\n- ISBN, for books\n- Degree type and institution, for dissertations\n- Standard number, for standards\n- URL and access date, for webpages when relevant\n\nThe audit table must include enough bibliographic information for the user to identify the reference without going back to the original list.\n\n## Verification Workflow\n\nFor each reference:\n\n1. Parse the submitted reference into structured fields:\n - reference number\n - authors\n - original title\n - translated title, if present\n - source / journal / book / conference / institution / issuing body\n - year\n - volume\n - issue\n - pages or article number\n - DOI\n - PMID / PMCID, if present\n - ISBN, if present\n - dissertation institution and degree type, if present\n - standard number, if present\n - URL and access date, if present\n\n2. Identify the source language and source type:\n - English-language journal article\n - Chinese-language journal article\n - bilingual Chinese journal article\n - dissertation / thesis\n - book / book chapter\n - conference paper\n - standard / guideline / policy / law / regulation\n - report / white paper\n - preprint\n - webpage / online document\n - dataset / software / patent / other\n\n3. Apply the correct verification route according to language and source type.\n\n For English-language or international journal references:\n - Verify by persistent identifier first when available.\n - DOI: resolve DOI and compare metadata.\n - PMID/PMCID: check PubMed / PubMed Central metadata when applicable.\n - ISBN: verify book metadata using publisher catalogue, library catalogue, or ISBN database.\n - Standard number: verify using the official standards platform or issuing body.\n - Patent number: verify using the official patent database.\n - If an identifier resolves but points to a different title, author, source, or year, mark as mismatch.\n - If an identifier does not resolve, mark as invalid identifier or Manual check depending on source type and available evidence.\n\n For Chinese-language references:\n - Do not use DOI-first verification by default.\n - Search the original Chinese title first in Chinese scholarly databases or official Chinese sources.\n - For Chinese-language journal articles, search the original Chinese title first using CNKI, Wanfang, VIP, official journal pages, and the publisher / society website when available.\n - For bilingual Chinese references, search the Chinese title first, then use the English translated title only as a secondary route.\n - For Chinese dissertations, search by Chinese title + author + institution using CNKI 博硕士论文库, Wanfang dissertations, university repository, National Library records, or institutional repository.\n - For Chinese books, search by Chinese title + author/editor + publisher + ISBN using publisher catalogues, National Library records, university library catalogues, or ISBN databases.\n - For Chinese standards, policies, laws, and regulations, search by standard number / document number / issuing authority using official government, standards, ministry, or institutional websites.\n - If DOI is included in a Chinese reference, check it only as a supplementary consistency check after the Chinese-title/source route, unless the target journal explicitly requires DOI verification.\n - Do not mark a Chinese reference as Major merely because DOI is absent.\n\n4. DOI handling rules.\n - For English-language or international journal articles, verify DOI metadata first when DOI is provided.\n - For Chinese-language references, DOI is optional unless the user, target journal, or citation style explicitly requires it.\n - If a Chinese reference has no DOI but is verified by CNKI, Wanfang, VIP, the official journal page, or another reliable Chinese source, it can be marked Verified.\n - If a Chinese reference includes a DOI and that DOI points to a different article, mark as Critical unless a clear typographic error explains it.\n - If DOI metadata conflicts with Chinese database metadata for a Chinese article, prefer the official journal page when available; otherwise record the conflict and mark confidence accordingly.\n\n5. If the primary route fails, broaden by title and source.\n - For biomedical English references, prefer PubMed, DOI metadata, publisher records, and official journal pages.\n - For general English scholarly references, prefer Crossref, publisher pages, official journal pages, institutional repositories, or library catalogues.\n - For Chinese references, first vary Chinese-title searches by removing punctuation, adding first author, adding journal/source, adding year, or trying simplified/traditional variants. Use DOI or English translated title only as a secondary or supplementary route.\n\n6. Compare submitted metadata against verified metadata:\n - title wording\n - translated title, if present\n - authors\n - journal/source\n - year\n - volume\n - issue\n - pages/article number\n - DOI\n - PMID/PMCID\n - Chinese database traceability\n - ISBN / standard number / document number\n\n7. Assign a match quality:\n - Exact: all key identity fields match.\n - Near exact: identity is clear; only minor formatting, punctuation, translation, capitalization, or abbreviation differences exist.\n - Partial: identity is likely but one or more important fields are missing or inconsistent.\n - Mismatch: identifier or metadata points to a different source.\n - Not found: no reliable trace found after reasonable identifier and title/source searches.\n - Ambiguous: multiple plausible records or conflicting metadata.\n\n8. Assign a status and severity:\n - Verified: key fields match reliable metadata.\n - Minor: formatting issue, capitalization issue, journal abbreviation inconsistency, missing issue, Chinese/English punctuation inconsistency, small non-critical discrepancy, or non-identity-changing title translation difference.\n - Major: wrong year, wrong title wording that changes identity, wrong journal, wrong author, wrong volume/pages, missing DOI for a recent English/international article where DOI is clearly available and required by the target style, missing Chinese source information that prevents traceability, or citation of a preprint when a peer-reviewed version clearly exists. For Chinese references, absence of DOI alone is not a Major issue.\n - Critical: likely fake, non-existent, DOI points to a different article, severe metadata conflict, fabricated PMID/DOI/CNKI-like identifier, or citation points to an entirely different source.\n - Manual check: ambiguous source, insufficient metadata, paywalled or inaccessible metadata, conflicting databases, non-indexed local source, print-only source, or source that requires manual database access.\n\n9. Check whole-list issues:\n - duplicate references\n - inconsistent journal abbreviations\n - inconsistent Chinese journal names, such as mixing full names and abbreviations\n - inconsistent DOI format\n - inconsistent Chinese citation punctuation and spacing\n - inconsistent bilingual title handling\n - inconsistent author-name format, especially Chinese names in pinyin vs Chinese characters\n - preprint cited when peer-reviewed version exists\n - possible retracted article\n - missing required fields\n - mismatch between in-text citations and reference list, if manuscript text is provided\n\n## Chinese-Language Reference Verification\n\nChinese references are verifiable even when they do not have DOI, PMID, or English metadata. Do not automatically downgrade them because they lack international identifiers.\n\nFor Chinese references following GB/T 7714 or common Chinese journal reference styles, DOI is normally optional unless the target journal explicitly asks for it. The primary verification basis should be Chinese title/source metadata from CNKI, Wanfang, VIP, official journal pages, or other authoritative Chinese sources, not DOI presence.\n\n### Preferred Search Order for Chinese Journal Articles\n\nUse the following route whenever possible:\n\n1. Exact Chinese title search in CNKI.\n2. Exact Chinese title search in Wanfang Data.\n3. Exact Chinese title search in VIP / CQVIP.\n4. Official journal website or journal sponsor / publisher page.\n5. University, society, institutional, or repository page.\n6. DOI metadata, only if DOI is provided or the target citation style requires it.\n7. General web title search as supporting evidence only.\n\nIf CNKI and Wanfang disagree on minor formatting, prefer the official journal page when available. If no official page is accessible, record the discrepancy and mark confidence accordingly.\n\n### Chinese Title Search Rules\n\nWhen searching Chinese references:\n\n- Search the original Chinese title first.\n- Use exact-title search whenever possible.\n- Remove only obvious punctuation differences if exact search fails.\n- Try simplified and traditional Chinese variants if needed.\n- Search title + first author when the title is short or generic.\n- Search title + journal name when many irrelevant records appear.\n- Search title + year when title and journal are insufficient.\n- If an English translated title is provided, use it only after checking the Chinese title.\n- Do not treat different English translations of the same Chinese title as major errors unless the source identity changes.\n\n### CNKI / Wanfang / VIP Evidence Handling\n\nWhen a Chinese reference is verified through Chinese databases, record the database route clearly:\n\n- CNKI title search\n- Wanfang title search\n- VIP title search\n- Official journal page\n- Chinese database + official journal page\n- Manual check: database access required\n\nIf the database page cannot be fully opened but search-result metadata is sufficient to identify the source, mark as Partial or Near exact depending on field completeness, and set confidence to Medium unless the official journal page confirms it.\n\n### Chinese Journal Articles\n\nFor Chinese journal articles, verify:\n\n- Chinese title\n- Authors in Chinese characters\n- Journal full Chinese name\n- Year\n- Volume, if available\n- Issue\n- Page range or article number\n- DOI, if available\n- Fund/project or abstract information only if needed to disambiguate records\n\nCommon issues to flag:\n\n- Wrong journal Chinese name\n- Missing issue number when the journal uses issue-based pagination\n- Wrong page range\n- Wrong publication year\n- Confusion between online-first year and issue year\n- Translated English title treated as official title without Chinese title\n- Inconsistent pinyin author names when the target journal requires English references\n\n### Chinese Dissertations / Theses\n\nFor Chinese dissertations, verify:\n\n- Author\n- Chinese title\n- Degree type: 硕士 / 博士 / 专业学位, if available\n- Institution\n- Year\n- Advisor, if available and required by citation style\n- Database or repository: CNKI, Wanfang, university repository, National Library, or institutional record\n\nIf a dissertation appears only in a university repository or only in CNKI/Wanfang, it can still be verified. Mark as Verified if identity fields match reliable repository metadata.\n\n### Chinese Books and Book Chapters\n\nFor Chinese books, verify:\n\n- Author/editor/translator\n- Chinese title\n- Publisher\n- Place of publication, if required by citation style\n- Edition\n- Year\n- ISBN\n- Cited page range, if provided\n\nPreferred verification routes:\n\n1. Publisher catalogue\n2. National Library / university library catalogue\n3. ISBN database\n4. Major library catalogue\n5. Book sales pages only as supporting evidence, not primary evidence\n\nFor book chapters, additionally verify chapter title, chapter author, editor, page range, and book title.\n\n### Chinese Conference Papers\n\nFor Chinese conference papers, verify:\n\n- Paper title\n- Authors\n- Conference name\n- Organizer or proceedings title\n- Location, if provided\n- Date/year\n- Page range, if available\n- Database source, such as CNKI conference proceedings or Wanfang conference papers\n\nIf only an abstract or meeting report is found, mark as Partial or Manual check depending on citation style requirements.\n\n### Chinese Standards, Laws, Policies, and Government Documents\n\nFor standards, policies, laws, regulations, and official documents, verify using official sources whenever possible.\n\nVerify:\n\n- Standard/document number\n- Full title\n- Issuing body\n- Release date\n- Implementation/effective date, if relevant\n- Version or replacement status\n- Whether the standard/document is current, superseded, abolished, or replaced\n- Official URL or official database route\n\nPreferred routes:\n\n1. National standards full-text disclosure system or official standards platform\n2. Official ministry / commission / government website\n3. Official gazette or government bulletin\n4. Issuing organization website\n5. Standards databases only as supporting evidence if official route is inaccessible\n\nFlag superseded or abolished standards prominently.\n\n### Chinese Patents\n\nFor Chinese patents, verify:\n\n- Patent number / application number\n- Title\n- Inventors\n- Applicants / assignees\n- Application date\n- Publication date\n- Grant status, if relevant\n- Patent type: invention, utility model, design\n- Official route: CNIPA or official patent database\n\nMark as Major if the patent status, applicant, or title is wrong. Mark as Critical if the number points to a different patent.\n\n## English-Language Reference Verification\n\nFor English references, use the following preferred routes:\n\n1. DOI metadata\n2. PubMed / PubMed Central, for biomedical literature\n3. Publisher page\n4. Crossref\n5. Official journal page\n6. Institutional repository\n7. Library catalogue, for books\n8. General web search only as supporting evidence\n\nFor biomedical references, PubMed, DOI metadata, and publisher records are preferred over secondary databases.\n\n## Retractions, Expressions of Concern, and Source-Type Risk\n\nFor each scholarly reference, check for visible retraction or expression-of-concern signals when feasible.\n\nFlag:\n\n- Retracted articles\n- Expressions of concern\n- Withdrawn preprints\n- Preprints where a peer-reviewed version exists\n- Predatory or non-indexed journals when source legitimacy is relevant\n- News, blogs, or webpages cited as scholarly evidence\n- Chinese webpages or policy pages that lack stable URL, issuing authority, or release date\n\nDo not overstate retraction status unless a reliable source confirms it.\n\n## Required Per-Round Output\n\nEvery response must include a per-reference audit table for the references checked in that round.\n\nUse this table format:\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | Identifier / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n\nRules for the table:\n\n- Do not omit title, author, or journal/source columns.\n- Use short but identifiable titles if titles are very long.\n- Use first author et al. if the author list is long.\n- Put DOI, PMID/PMCID, ISBN, standard number, patent number, CNKI/Wanfang/VIP route, or official-source route in the same identifier field as appropriate.\n- For Chinese references, prioritize CNKI/Wanfang/VIP/official-source trace in this field. Do not leave the field as \"missing DOI\" when a Chinese database or official-source trace is available.\n- Verification Route should state how it was checked, such as DOI metadata, PubMed, publisher page, Crossref, CNKI title search, Wanfang title search, VIP title search, official journal page, university repository, publisher catalogue, official standards website, or manual confirmation needed.\n- Match Quality must be one of: Exact, Near exact, Partial, Mismatch, Not found, Ambiguous.\n- Status must be one of: Verified, Minor, Major, Critical, Manual check.\n- Confidence must be High, Medium, or Low.\n- For Chinese references, write the original Chinese title whenever available.\n- If the verified source uses an alternate English translation, mention it only when it matters.\n\n## Per-Round Closing Requirement\n\nAt the end of every response, state exactly which reference numbers were checked in this round.\n\nIf references remain unchecked, end with a continuation prompt:\n\n\"This round checked references X–Y. References Z–N remain unchecked. Would you like me to continue with the next round, references Z–...?\"\n\nDo not imply that the full audit is complete if references remain unchecked.\n\n## Final Completion Requirement\n\nWhen all references have been checked, state clearly:\n\n\"Reference audit complete.\"\n\nThen provide a cumulative summary of all problematic references found across the entire audit, grouped by severity:\n\n### Critical Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Major Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Minor Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Manual Checks Needed\n| Ref | Title | Authors | Source / Journal | Reason | Suggested Manual Search |\n\nIf no problems are found in a category, write \"None identified.\"\n\n## Output Style\n\nBe concise but complete. Prioritize traceable verification over polished prose. Never fabricate metadata. If uncertain, say so clearly and mark the reference as Manual check.\n\nFor Chinese references, prefer clear Chinese wording in the issue/fix column. Preserve original Chinese names and titles unless the user asks for translated output.\n\n## Important Constraints\n\n- Do not invent DOI, PMID, PMCID, CNKI identifiers, Wanfang identifiers, VIP identifiers, ISBNs, page numbers, authors, journal names, or official document numbers.\n- Do not mark a reference as verified unless a reliable verification route supports it.\n- For biomedical references, prefer PubMed, DOI metadata, and publisher records.\n- For Chinese-language references, prefer Chinese-title-first verification through CNKI, Wanfang, VIP, official journal pages, university repositories, publisher catalogues, official standards platforms, official government websites, and CNIPA depending on source type. DOI is supplementary and optional unless explicitly required.\n- Distinguish article numbers from page ranges.\n- Distinguish online publication year from issue publication year when relevant.\n- Distinguish Chinese original titles from English translated titles.\n- Do not treat translation variation as a major error unless the source identity changes.\n- Label preprints clearly.\n- Flag retracted articles prominently when detected.\n- Flag superseded standards, abolished documents, and withdrawn policies when detected.\n- If database access is unavailable or paywalled, mark as Manual check or Partial rather than fabricating confirmation.\n\nFile v1.0.1:README.md\n\n
\n\n# Reference Checker Skill｜参考文献核查 Skill\n\n🌐 Language: 中文 | [English](README.md)\n\n**不要让一条假文献毁掉整篇论文的可信度。**\n\nReference Checker Skill 是一个用于**学术论文投稿前参考文献逐条核查**的 AI Skill，适合研究生、科研作者、编辑、审稿助理和学术写作者使用。\n\n它的目标不是“帮你简单看一下参考文献”，而是让 AI 按照严格流程，对参考文献列表进行**逐条、结构化、可追踪的核验**，尽量发现投稿前最容易被忽略、但最可能影响论文可信度的问题。\n\nv1.4.0 起，本 Skill 明确支持**英文文献与中文文献的双语核查**。除 DOI、PubMed、Crossref、出版社页面等常规路径外，中文文献会优先使用**中文原题名**进行核查，并将**知网 CNKI、万方 Wanfang、维普 VIP/CQVIP、期刊官网、出版社目录、大学机构库、国家标准/政府官网**等纳入检索路径。\n\n---\n\n## v1.4.0 更新重点\n\n这一版主要增强了中文文献核查能力：\n\n* **支持中文文献逐条检查**：中文期刊论文、中文学位论文、中文图书、中文会议论文、标准、政策文件、专利等均可纳入核查。\n* **增加中文数据库检索路径**：中文期刊优先按“中文原题名”检索，推荐依次核对期刊官网、知网、万方、维普等来源。\n* **避免误判无 DOI 中文文献**：很多中文文献没有 DOI、PMID 或英文元数据，Skill 会要求 AI 不得仅因缺少这些字段就判定为不可验证。\n* **增加中文溯源字段**：输出表格中的 `DOI / PMID` 扩展为 `DOI / PMID / Chinese Trace`，便于记录 CNKI、万方、维普、期刊官网、ISBN、标准号、专利号等信息。\n* **增强中文格式检查**：新增对中文期刊名、中文作者名、GB/T 7714 风格、中文标点、页码、卷期、学位授予单位、出版社、ISBN 等字段的核查要求。\n* **区分中文原题名与英文译名**：对于双语中文文献，应优先核验中文题名；英文译名不一致时，只有在影响文献身份识别时才标记为严重问题。\n\n---\n\n## 为什么需要这个 Skill？\n\n在 AI 辅助写作越来越普遍的今天，参考文献错误变得更隐蔽，也更危险。\n\n很多错误看起来“很像真的”：\n\n* DOI 存在，但指向的是另一篇文章\n* 题名看起来合理，但数据库里查不到\n* 作者、期刊、年份、卷期页码彼此错配\n* 中文题名、英文译名、期刊英文名之间不一致\n* 中文文献没有 DOI，被错误地当作“无法核验”\n* 预印本已经有正式发表版本，但文中仍引用旧版本\n* 同一篇文献被重复引用\n* AI 生成了格式完整、内容逼真、但实际不存在的参考文献\n\n这些问题如果出现在投稿稿件中，轻则影响编辑和审稿人的第一印象，重则直接损害整篇论文的可信度。\n\n**Reference Checker Skill 的核心价值，就是在投稿前帮你把这些问题尽可能提前暴露出来。**\n\n---\n\n## 这个 Skill 能做什么？\n\nReference Checker Skill 会要求 AI 对参考文献进行**逐条核查**，而不是只挑几条看起来可疑的文献抽样检查。\n\n它会尽量核验每一条参考文献的关键信息，包括：\n\n| 检查项目 | 是否检查 |\n| --- | --- |\n| 文献是否真实存在 | 是 |\n| 文章题名 / 中文原题名 | 是 |\n| 作者信息 / 中文作者名 | 是 |\n| 期刊 / 来源 / 出版社 / 授予单位 | 是 |\n| 发表年份 / 出版年份 / 授予年份 | 是 |\n| 卷、期、页码或文章编号 | 是 |\n| DOI | 如有则检查 |\n| PMID / PMCID | 如有则检查 |\n| CNKI / 万方 / 维普 / 期刊官网溯源 | 中文文献优先检查 |\n| ISBN / 标准号 / 专利号 / 文件号 | 对应文献类型检查 |\n| 重复引用 | 是 |\n| 预印本与正式发表版本冲突 | 是 |\n| 撤稿或来源风险 | 可检测时检查 |\n| 中文/英文题名不一致 | 是 |\n| GB/T 7714 或目标期刊格式一致性 | 可按需求检查 |\n\n---\n\n## 核心特点\n\n### 1. 逐条核查，而不是抽样检查\n\n很多参考文献检查工具只适合快速浏览，容易漏掉隐藏错误。\n\n这个 Skill 默认要求 AI **逐条检查每一条参考文献**。除非用户明确要求抽样，否则不应跳过任何一条。\n\n---\n\n### 2. 不只查 DOI，还查元数据是否匹配\n\n一个 DOI 能打开，并不代表这条参考文献就是正确的。\n\n这个 Skill 会要求 AI 同时核对：\n\n* 题名是否一致\n* 作者是否匹配\n* 期刊名称是否正确\n* 年份、卷期、页码是否对应\n* DOI 是否真的指向该文章\n* PMID / PMCID 是否与文献信息一致\n* 中文文献是否能通过知网、万方、维普、期刊官网或其他可靠来源溯源\n\n它关注的不是“有没有 DOI”，而是**整条参考文献是否真实、准确、对应一致**。\n\n---\n\n### 3. 支持中文文献数据库检索\n\n对于中文文献，Skill 会要求 AI 优先按**中文原题名**核查，而不是直接用英文译名检索。\n\n推荐检索路径包括：\n\n| 文献类型 | 推荐核查路径 |\n| --- | --- |\n| 中文期刊论文 | 期刊官网、知网 CNKI、万方、维普、DOI 元数据 |\n| 中文学位论文 | 知网博硕士论文库、万方学位论文库、学校机构库、国家图书馆记录 |\n| 中文图书 / 章节 | 出版社目录、ISBN 数据库、国家图书馆、高校图书馆目录 |\n| 中文会议论文 | 会议官网、论文集、知网会议库、万方会议库 |\n| 中文标准 | 国家标准全文公开系统、行业标准官网、主管部门网站 |\n| 政策 / 法律 / 政府文件 | 官方政府网站、发文机关网站、法规数据库 |\n| 中国专利 | 国家知识产权局、专利公布公告系统 |\n\n如果中文数据库之间存在细微差异，Skill 会要求记录差异、说明核查路径，并根据证据强弱给出置信度，而不是简单给出“已验证”或“未找到”。\n\n---\n\n### 4. 对问题进行严重程度分级\n\n所有发现的问题都会被标记为不同严重程度，方便你优先处理。\n\n| 严重程度 | 含义 |\n| --- | --- |\n| `Critical` | 高风险问题，例如疑似假文献、不存在的文章、错误 DOI、DOI 指向另一篇文章、伪造 PMID/DOI/CNKI-like 信息 |\n| `Major` | 真实文献，但关键元数据错误，例如题名、期刊、年份、卷期页码明显不匹配，或缺失导致无法溯源的中文来源信息 |\n| `Minor` | 格式、大小写、字段不完整、中文/英文标点、期刊名缩写、引用风格不一致等小问题 |\n| `Manual check` | 证据不足、数据库冲突、来源不可访问或文献类型复杂，需要人工进一步确认 |\n\n---\n\n### 5. 支持长参考文献列表分批检查\n\n如果参考文献数量很多，AI 一次无法可靠检查完，Skill 会要求 AI 分批处理，并明确说明：\n\n* 本轮检查了哪些编号\n* 哪些参考文献还没有检查\n* 是否需要继续下一批\n\n这样可以避免 AI 在长列表中漏查、跳查或假装已经全部检查完成。\n\n---\n\n### 6. 输出结果适合人工复核\n\n每一轮检查都会输出结构化表格，方便作者、导师、编辑或合作者快速定位问题。\n\n示例输出表格包括：\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n\n核查完成后，Skill 还会要求 AI 汇总所有问题文献，方便用户一次性修改。\n\n---\n\n## 适合哪些场景？\n\n这个 Skill 特别适合：\n\n* 期刊论文投稿前检查\n* 学位论文参考文献核查\n* 综述论文长参考文献列表检查\n* 生物医学和生命科学论文投稿\n* 人文社科论文中的中文文献核查\n* 中文核心、CSSCI、北大核心、科技核心等中文期刊投稿前检查\n* 使用 AI 辅助写作后检查引用真实性\n* 文献管理软件批量导出后的质量控制\n* 编辑部或课题组内部的投稿前质控流程\n* 检查疑似 AI 生成或 AI 修改过的参考文献\n\n---\n\n## 示例提示词\n\n你可以这样使用：\n\n```text\nPlease use the Reference Checker Skill to audit the following reference list.\n\nRequirements:\n1. Check every reference item by item.\n2. Do not sample.\n3. Verify title, authors, journal, year, volume, issue, pages, DOI, and PMID if available.\n4. For Chinese references, search the original Chinese title first and use CNKI, Wanfang, VIP, official journal pages, or other reliable Chinese sources when applicable.\n5. Classify problems as Critical, Major, Minor, or Manual check.\n6. If the list is too long, check it in batches and ask me whether to continue.\n```\n\n中文用户也可以直接这样说：\n\n```text\n请使用 Reference Checker Skill 帮我逐条核查下面的参考文献列表。\n\n要求：\n1. 必须逐条检查，不要抽样。\n2. 核查题名、作者、期刊、年份、卷、期、页码、DOI，如果有 PMID 也一并检查。\n3. 中文文献请优先用中文原题名检索，并尽量通过知网、万方、维普、期刊官网、出版社目录、学校机构库或官方网站核查。\n4. 不要因为中文文献没有 DOI 或 PMID 就直接判定为无法验证。\n5. 将问题分为 Critical、Major、Minor 和 Manual check。\n6. 如果参考文献太多，请分批检查，并明确告诉我本轮检查了哪些编号，还有哪些没有检查。\n7. 全部检查完成后，请汇总所有有问题的参考文献。\n```\n\n---\n\n## 推荐输出格式\n\n每一轮核查应包含如下表格：\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID / Chinese Trace | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |\n\n如果参考文献较多，本轮结束时应说明：\n\n```text\nThis round checked references 1–20. References 21–58 remain unchecked.\nWould you like me to continue with the next round, references 21–40?\n```\n\n全部核查完成后，应输出：\n\n```text\nReference audit complete.\n```\n\n并汇总所有存在 `Critical`、`Major`、`Minor` 和 `Manual check` 问题的参考文献。\n\n---\n\n## 推荐仓库结构\n\n```text\nreference-checker-skill/\n├── SKILL.md\n├── README.md\n├── README.zh-CN.md\n├── README.en.md\n├── assets/\n│ └── cover.png\n├── templates/\n│ ├── report_template.md\n│ └── correction_table_template.md\n└── examples/\n ├── input_references.txt\n ├── input_references_zh.txt\n └── expected_output.md\n```\n\n---\n\n## 安装与使用\n\n下载或克隆本仓库后，将文件夹放入支持 Skill 工作流的 AI 环境中。\n\n核心文件是：\n\n```text\nSKILL.md\n```\n\n如果你的环境不支持 Skill 文件夹，也可以直接将 `SKILL.md` 中的内容复制到：\n\n* 自定义指令\n* 项目指令\n* Agent 工作流\n* Manuscript checking prompt\n* 投稿前质控流程\n\n---\n\n## 使用建议\n\n为了提高核查准确性，建议用户提供尽可能完整的参考文献信息：\n\n* 对英文文献：题名、作者、期刊、年份、卷期页码、DOI、PMID/PMCID。\n* 对中文期刊论文：中文题名、中文作者、期刊全称、年份、卷期页码、DOI 或数据库来源。\n* 对中文学位论文：题名、作者、授予单位、学位类型、年份、数据库来源。\n* 对中文图书：书名、作者/编者、出版社、出版年份、ISBN、引用页码。\n* 对标准/政策/专利：标准号、文件号、专利号、发布机构、发布日期。\n\n---\n\n## 重要说明\n\nReference Checker Skill 用于辅助参考文献核查，但不能完全替代人工审阅。\n\n对于被标记为以下类型的参考文献，建议用户务必人工复核：\n\n* `Critical`\n* `Major`\n* `Manual check`\n\n另外，知网、万方、维普等中文数据库可能存在访问权限限制。Skill 的作用是要求 AI 采用正确的检索路径、记录证据来源和不确定性；如果数据库无法访问，应明确标记为 `Manual check`，而不是臆造核查结果。\n\n这个 Skill 的设计原则是：**宁可严格标记可疑问题，也不要把无法确认的参考文献默认为正确。**\n\n---\n\n## License\n\nMIT License.\n\nFile v1.0.1:_meta.json\n\n{\n \"ownerId\": \"kn72ayzheax209248fk1zdzfrx85h3ah\",\n \"slug\": \"reference-checker\",\n \"version\": \"1.0.1\",\n \"publishedAt\": 1780487792272\n}\n\nFile v1.0.1:skill-card.md\n\n## Description: \nExhaustively verifies English and Chinese manuscript references before journal submission for authenticity, metadata accuracy, traceability, and formatting consistency. \n\nThis skill is ready for commercial/non-commercial use. \n\n## Publisher: \n[liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai) \n\n### License/Terms of Use: \nMIT-0 \n\n\n## Use Case: \nResearchers, academic authors, editors, review assistants, and manuscript-preparation teams use this skill to audit every reference in a bibliography for existence, metadata accuracy, identifier traceability, duplicate entries, source risks, and citation-format consistency. \n\n### Deployment Geography for Use: \nGlobal \n\n## Known Risks and Mitigations: \nRisk: Reference audit findings may be incomplete or misleading when database access is unavailable, paywalled, or inconsistent across sources. \nMitigation: Manually review Critical, Major, and Manual check results, especially where the skill reports limited access, conflicting databases, or ambiguous records. \nRisk: The workflow may use web or scholarly database searches to validate manuscript references. \nMitigation: Use approved search sources for sensitive manuscripts and avoid sharing confidential material with services that are not authorized for the user's workflow. \n\n\n## Reference(s): \n- [ClawHub skill page](https://clawhub.ai/liuxiangjian-ai/reference-checker) \n\n\n## Skill Output: \n**Output Type(s):** [text, markdown, guidance] \n**Output Format:** [Markdown audit tables and summary text] \n**Output Parameters:** [1D] \n**Other Properties Related to Output:** [Per-reference audit rows with verification route, match quality, status, confidence, main issue, and suggested fix.] \n\n## Skill Version(s): \n1.0.1 (source: ClawHub release metadata) \n\n## Ethical Considerations: \nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. \n\nArchive v1.0.0: 8 files, 8685 bytes\n\nFiles: examples/expected_output.md (1297b), examples/input_references.txt (283b), README.md (5544b), skill-card.md (2402b), SKILL.md (6625b), templates/correction_table_template.md (177b), templates/report_template.md (1022b), _meta.json (136b)\n\nFile v1.0.0:SKILL.md\n\n---\nname: reference-checker\nversion: 1.3.0\ndescription: Exhaustively verify manuscript references before journal submission. Use when checking whether references are real, accurate, complete, and formatted consistently.\n---\n\n# Reference Checker Skill\n\n## Purpose\n\nYou are a pre-submission reference verification assistant. Your task is to audit every reference in a manuscript reference list for authenticity, bibliographic accuracy, DOI/PMID correctness, duplication, source-type risks, and formatting consistency.\n\nThis skill is designed for exhaustive reference checking, not sampling.\n\n## Default Mode: Exhaustive Item-by-Item Audit\n\nUnless the user explicitly requests sampling, you must check every reference one by one.\n\nYou must not:\n- Skip references because the list is long.\n- Only check suspicious-looking references.\n- Summarize without producing a per-reference audit table.\n- Mark a reference as verified based only on plausibility.\n- Collapse multiple references into one generic comment.\n- Stop without telling the user exactly which reference numbers have been checked and which remain unchecked.\n\nIf the reference list is too long for one response, process it in sequential batches. Continue from the last checked reference number in the next round.\n\n## Required Data Fields for Each Reference\n\nFor every reference, extract and display the following fields whenever available:\n\n- Reference number\n- Reference title\n- First author or submitted author string\n- Journal / book / conference / source\n- Year\n- DOI\n- PMID / PMCID, if available\n\nThe audit table must include enough bibliographic information for the user to identify the reference without going back to the original list.\n\n## Verification Workflow\n\nFor each reference:\n\n1. Parse the submitted reference into structured fields:\n - reference number\n - authors\n - title\n - source / journal\n - year\n - volume\n - issue\n - pages or article number\n - DOI\n - PMID / PMCID, if present\n\n2. Verify by DOI first when DOI is provided.\n - If the DOI resolves and metadata matches, mark as verified.\n - If the DOI resolves but points to a different title, author, journal, or year, mark as DOI mismatch.\n - If the DOI does not resolve, mark as invalid DOI or needs manual confirmation.\n\n3. If DOI is missing, invalid, or suspicious, verify by exact title.\n - For biomedical references, prefer PubMed, DOI metadata, publisher records, and official journal pages.\n - Crossref and publisher metadata are preferred for general scholarly literature.\n - Secondary databases may be used as supporting evidence but should not override DOI, PubMed, or publisher metadata.\n\n4. Compare submitted metadata against verified metadata:\n - title\n - authors\n - journal/source\n - year\n - volume\n - issue\n - pages/article number\n - DOI\n - PMID/PMCID\n\n5. Assign a status and severity:\n - Verified: key fields match reliable metadata.\n - Minor: formatting issue, capitalization issue, journal abbreviation inconsistency, missing issue, small non-critical discrepancy.\n - Major: wrong year, wrong title wording that changes identity, wrong journal, wrong author, wrong volume/pages, missing DOI for a recent article where DOI is clearly available.\n - Critical: likely fake, non-existent, DOI points to a different article, severe metadata conflict.\n - Manual check: ambiguous source, insufficient metadata, paywalled or inaccessible metadata, conflicting databases, preprint/published-version uncertainty.\n\n6. Check whole-list issues:\n - duplicate references\n - inconsistent journal abbreviations\n - inconsistent DOI format\n - preprint cited when peer-reviewed version exists\n - possible retracted article\n - missing required fields\n - mismatch between in-text citations and reference list, if manuscript text is provided\n\n## Required Per-Round Output\n\nEvery response must include a per-reference audit table for the references checked in that round.\n\nUse this table format:\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n\nRules for the table:\n\n- Do not omit title, author, or journal/source columns.\n- Use short but identifiable titles if titles are very long.\n- Use first author et al. if the author list is long.\n- Put DOI and PMID/PMCID in the same field if both exist.\n- Verification Route should state how it was checked, such as DOI metadata, PubMed, publisher page, Crossref, title search, or manual confirmation needed.\n- Match Quality should be one of: Exact, Near exact, Partial, Mismatch, Not found, Ambiguous.\n- Status should be one of: Verified, Minor, Major, Critical, Manual check.\n- Confidence should be High, Medium, or Low.\n\n## Per-Round Closing Requirement\n\nAt the end of every response, state exactly which reference numbers were checked in this round.\n\nIf references remain unchecked, end with a continuation prompt:\n\n\"This round checked references X–Y. References Z–N remain unchecked. Would you like me to continue with the next round, references Z–...?\"\n\nDo not imply that the full audit is complete if references remain unchecked.\n\n## Final Completion Requirement\n\nWhen all references have been checked, state clearly:\n\n\"Reference audit complete.\"\n\nThen provide a cumulative summary of all problematic references found across the entire audit, grouped by severity:\n\n### Critical Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Major Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Minor Problems\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n\n### Manual Checks Needed\n| Ref | Title | Authors | Source / Journal | Reason | Suggested Manual Search |\n\nIf no problems are found in a category, write \"None identified.\"\n\n## Output Style\n\nBe concise but complete. Prioritize traceable verification over polished prose. Never fabricate metadata. If uncertain, say so clearly and mark the reference as Manual check.\n\n## Important Constraints\n\n- Do not invent DOI, PMID, PMCID, page numbers, authors, or journal names.\n- Do not mark a reference as verified unless a reliable verification route supports it.\n- For biomedical references, prefer PubMed, DOI metadata, and publisher records.\n- Distinguish article numbers from page ranges.\n- Distinguish online publication year from issue publication year when relevant.\n- Label preprints clearly.\n- Flag retracted articles prominently when detected.\n\nFile v1.0.0:README.md\n\n# Reference Checker Skill\n\n**An exhaustive pre-submission reference verification skill for scholarly manuscripts.**\n\nReference Checker Skill helps researchers, students, editors, and academic writers audit manuscript references before journal submission. \nIt is designed to reduce fake citations, DOI mismatches, incorrect metadata, duplicated references, and AI-generated reference errors.\n\nUnlike lightweight citation checkers that only sample a few references, this skill is designed for **item-by-item exhaustive verification**.\n\n---\n\n## Why this skill?\n\nReference errors are easy to miss but costly during submission.\n\nCommon problems include:\n\n- fabricated or non-existent references\n- DOI pointing to the wrong article\n- incorrect article titles\n- wrong journal names\n- mismatched year, volume, issue, or page numbers\n- duplicated references\n- preprints cited when a peer-reviewed version already exists\n- inconsistent reference formatting\n- AI-generated references that look plausible but are not real\n\nThis skill forces the AI to audit **every reference**, not just the suspicious ones.\n\n---\n\n## Key Features\n\n- **Exhaustive reference checking** \n Every reference must be reviewed individually unless the user explicitly requests sampling.\n\n- **Structured metadata verification** \n Checks title, authors, journal/source, year, DOI, PMID, volume, issue, and pages.\n\n- **Batch-by-batch workflow for long reference lists** \n If the reference list is too long, the skill reports which references were checked in the current round and asks whether to continue.\n\n- **Clear severity labels** \n Problems are classified as:\n - `Critical`\n - `Major`\n - `Minor`\n - `Manual check`\n\n- **Final problem summary** \n When the full audit is complete, the skill summarizes all problematic references in one place.\n\n- **Human-review friendly tables** \n Each output table includes the submitted title, authors, journal/source, year, DOI/PMID, verification route, match quality, status, confidence, and suggested fix.\n\n- **No Python dependency** \n This is a pure prompt-based skill. No scripts or external setup are required.\n\n---\n\n## What it checks\n\nFor each reference, the skill attempts to verify:\n\n| Field | Checked |\n|---|---|\n| Reference existence | Yes |\n| Article title | Yes |\n| Authors | Yes |\n| Journal / source | Yes |\n| Publication year | Yes |\n| Volume / issue | Yes |\n| Page range / article number | Yes |\n| DOI | Yes |\n| PMID / PMCID | When available |\n| Duplicate references | Yes |\n| Preprint vs published version conflict | Yes |\n| Retraction or source risk | When detectable |\n\n---\n\n## Recommended Use Cases\n\nThis skill is especially useful for:\n\n- journal manuscript submission\n- review articles with long reference lists\n- biomedical and life science manuscripts\n- AI-assisted academic writing\n- reference lists reformatted by citation managers\n- checking references generated or modified by LLMs\n- pre-submission editorial quality control\n\n---\n\n## Example Prompt\n\n```text\nPlease use the Reference Checker Skill to audit the following reference list.\n\nRequirements:\n1. Check every reference item by item.\n2. Do not sample.\n3. Verify title, authors, journal, year, volume, issue, pages, DOI, and PMID if available.\n4. Classify problems as Critical, Major, Minor, or Manual check.\n5. If the list is too long, check it in batches and ask me whether to continue.\n```\n\n---\n\n## Output Format\n\nEach audit round includes a table like this:\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n\nFor long reference lists, the skill ends each round with a continuation message, for example:\n\n```text\nThis round checked references 1–20. References 21–58 remain unchecked.\nWould you like me to continue with the next round, references 21–40?\n```\n\nWhen all references have been checked, the skill provides a final summary:\n\n```text\nReference audit complete.\n```\n\nThen it lists all references with Critical, Major, Minor, and Manual check issues.\n\n---\n\n## Severity Definitions\n\n| Severity | Meaning |\n|---|---|\n| Critical | Likely fake reference, non-existent article, wrong DOI, or DOI points to a different article |\n| Major | Real article, but important metadata is wrong |\n| Minor | Formatting problem, incomplete field, capitalization issue, or style inconsistency |\n| Manual check | Not enough evidence to verify confidently |\n\n---\n\n## Repository Structure\n\n```text\nreference-checker-skill/\n├── SKILL.md\n├── README.md\n├── templates/\n│ ├── report_template.md\n│ └── correction_table_template.md\n└── examples/\n ├── input_references.txt\n └── expected_output.md\n```\n\n---\n\n## Installation\n\nDownload or clone this repository, then place the folder in your AI skill directory if your environment supports skill-based workflows.\n\nThe core file is:\n\n```text\nSKILL.md\n```\n\nYou can also copy the contents of `SKILL.md` into a custom instruction, project instruction, or agent workflow.\n\n---\n\n## Important Note\n\nThis skill is designed to support reference verification, not to replace final human review.\n\nFor journal submission, users should manually confirm all references marked as:\n\n- `Critical`\n- `Major`\n- `Manual check`\n\nThe skill is intentionally strict. If a reference cannot be confidently verified, it should be flagged rather than silently accepted.\n\n---\n\n## License\n\nMIT License.\n\nFile v1.0.0:_meta.json\n\n{\n \"ownerId\": \"kn72ayzheax209248fk1zdzfrx85h3ah\",\n \"slug\": \"reference-checker\",\n \"version\": \"1.0.0\",\n \"publishedAt\": 1777121632762\n}\n\nFile v1.0.0:examples/expected_output.md\n\n# Reference Verification Report\n\n## This Round\n\nReferences checked in this round: 1–2\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n| 1 | Artificial intelligence in lyophilization process monitoring | Smith J, Lee K | J Pharm Sci | 2021 | 10.0000/example-doi | DOI metadata / title search needed | Not found | Manual check | Low | DOI appears invalid in this example; verify by exact title and publisher search. |\n| 2 | Freeze drying of biologics: challenges and opportunities | Wang X, Chen Y | Pharmaceutics | 2022 | 10.3390/pharmaceutics14050999 | DOI metadata | Exact | Verified | High | No issue identified in this example. |\n\n## Problems Identified in This Round\n\n### Critical\nNone identified.\n\n### Major\nNone identified.\n\n### Minor\nNone identified.\n\n### Manual Check\n| Ref | Title | Authors | Source / Journal | Reason | Suggested Manual Search |\n|---|---|---|---|---|---|\n| 1 | Artificial intelligence in lyophilization process monitoring | Smith J, Lee K | J Pharm Sci | DOI appears invalid in this example. | Search exact title in Crossref, PubMed, and publisher website. |\n\nReference audit complete.\n\nFile v1.0.0:skill-card.md\n\n## Description: \nExhaustively verify manuscript references before journal submission. Use when checking whether references are real, accurate, complete, and formatted consistently. \n\nThis skill is ready for commercial/non-commercial use. \n\n## Publisher: \n[liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai) \n\n### License/Terms of Use: \nMIT-0 \n\n\n## Use Case: \nResearchers, students, editors, and academic writers use this skill to audit manuscript reference lists before journal submission. It checks each citation for existence, metadata accuracy, DOI/PMID correctness, duplicates, source-type risks, and formatting consistency. \n\n### Deployment Geography for Use: \nGlobal \n\n## Known Risks and Mitigations: \nRisk: Confidential manuscripts or unpublished reference lists may be exposed if an agent performs external DOI, PubMed, Crossref, or publisher lookups. \nMitigation: Provide only the reference list when possible, or confirm that external bibliographic lookups are acceptable before using the skill on confidential material. \nRisk: Bibliographic metadata may be inaccessible, ambiguous, paywalled, conflicting, or outdated, which can lead to uncertain verification results. \nMitigation: Treat Critical, Major, and Manual check findings as requiring human review against primary DOI, PubMed, publisher, or journal records. \n\n\n## Reference(s): \n- [ClawHub skill listing](https://clawhub.ai/liuxiangjian-ai/reference-checker) \n- [Publisher profile](https://clawhub.ai/user/liuxiangjian-ai) \n- [README](artifact/README.md) \n- [Skill instructions](artifact/SKILL.md) \n\n\n## Skill Output: \n**Output Type(s):** [Analysis, Markdown, Guidance] \n**Output Format:** [Markdown tables and concise prose] \n**Output Parameters:** [1D] \n**Other Properties Related to Output:** [Produces per-reference audit rows, severity groupings, confidence labels, suggested fixes, and continuation status for long reference lists.] \n\n## Skill Version(s): \n1.0.0 (source: server release metadata; artifact frontmatter says 1.3.0) \n\n## Ethical Considerations: \nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. \n\nFile v1.0.0:templates/correction_table_template.md\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | Current Field | Verified / Suggested Correction | Severity |\n|---|---|---|---|---|---|---|---|\n\nFile v1.0.0:templates/report_template.md\n\n# Reference Verification Report\n\n## This Round\n\nReferences checked in this round: [X–Y]\n\n| Ref | Submitted Title | Submitted Authors | Submitted Source / Journal | Year | DOI / PMID | Verification Route | Match Quality | Status | Confidence | Main Issue / Suggested Fix |\n|---|---|---|---|---|---|---|---|---|---|---|\n| 1 | | | | | | | | | | |\n\n## Problems Identified in This Round\n\n### Critical\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n|---|---|---|---|---|---|\n\n### Major\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n|---|---|---|---|---|---|\n\n### Minor\n| Ref | Title | Authors | Source / Journal | Problem | Suggested Action |\n|---|---|---|---|---|---|\n\n### Manual Check\n| Ref | Title | Authors | Source / Journal | Reason | Suggested Manual Search |\n|---|---|---|---|---|---|\n\n## Continuation Status\n\nThis round checked references X–Y. References Z–N remain unchecked. Would you like me to continue with the next round, references Z–...?\n\nFile v1.0.0:examples/input_references.txt\n\n1. Smith J, Lee K. Artificial intelligence in lyophilization process monitoring. J Pharm Sci. 2021;110(4):1234-1245. doi:10.0000/example-doi\n2. Wang X, Chen Y. Freeze drying of biologics: challenges and opportunities. Pharmaceutics. 2022;14(5):999. doi:10.3390/pharmaceutics14050999","readmeExcerpt":"Skill: Reference Checker Skill Owner: liuxiangjian-ai Summary: Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, t... Tags: academic-writing:1.0.0, ai-writing:1.0.0, citation-verification:1.0.0, crossref:1.0.0, doi:1.0.0, journal-submission:1.0.0, latest:1.0.2, manuscript-review:1.0.0, pubmed:1.0.0, reference-che","codeSnippets":[],"executableExamples":[{"language":"text","snippet":"Please use the Reference Checker Skill to audit the following reference list.\n\nRequirements:\n1. Check every reference item by item.\n2. Do not sample.\n3. Verify title, authors, journal, year, volume, issue, pages, DOI, and PMID if available.\n4. For Chinese references, search the original Chinese title first and use CNKI, Wanfang, VIP, official journal pages, or other reliable Chinese sources when applicable.\n5. Classify problems as Critical, Major, Minor, or Manual check.\n6. If the list is too long, check it in batches and ask me whether to continue."},{"language":"text","snippet":"请使用 Reference Checker Skill 帮我逐条核查下面的参考文献列表。\n\n要求：\n1. 必须逐条检查，不要抽样。\n2. 核查题名、作者、期刊、年份、卷、期、页码、DOI，如果有 PMID 也一并检查。\n3. 中文文献请优先用中文原题名检索，并尽量通过知网、万方、维普、期刊官网、出版社目录、学校机构库或官方网站核查。\n4. 不要因为中文文献没有 DOI 或 PMID 就直接判定为无法验证。\n5. 将问题分为 Critical、Major、Minor 和 Manual check。\n6. 如果参考文献太多，请分批检查，并明确告诉我本轮检查了哪些编号，还有哪些没有检查。\n7. 全部检查完成后，请汇总所有有问题的参考文献。"},{"language":"text","snippet":"This round checked references 1–20. References 21–58 remain unchecked.\nWould you like me to continue with the next round, references 21–40?"},{"language":"text","snippet":"Reference audit complete."},{"language":"text","snippet":"reference-checker-skill/\n├── SKILL.md\n├── README.md\n├── README.zh-CN.md\n├── README.en.md\n├── assets/\n│ └── cover.png\n├── templates/\n│ ├── report_template.md\n│ └── correction_table_template.md\n└── examples/\n ├── input_references.txt\n ├── input_references_zh.txt\n └── expected_output.md"},{"language":"text","snippet":"SKILL.md"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: reference-checker\nversion: 1.4.1\ndescription: Exhaustively verify English and Chinese manuscript references before journal submission. Use when checking whether references are real, accurate, complete, traceable, and formatted consistently. Uses DOI/PubMed/Crossref/publisher checks for English/international references, and uses Chinese-title-first CNKI/Wanfang/VIP/official-source checks for Chinese references. DOI is optional for Chinese references and must not be required unless the target citation style explicitly requires it.\n---\n\n# Reference Checker Skill\n\n## Purpose\n\nYou are a pre-submission reference verification assistant. Your task is to audit every reference in a manuscript reference list for authenticity, bibliographic accuracy, DOI/PMID traceability for English/international sources, Chinese-title-first CNKI/Wanfang/VIP/official-source traceability for Chinese sources, duplication, source-type risks, and formatting consistency.\n\nThis skill is designed for exhaustive reference checking, not sampling.\n\nIt supports both English-language and Chinese-language references, including journal articles, books, dissertations, conference papers, standards, policies, reports, preprints, datasets, webpages, and other citable sources.\n\n## Default Mode: Exhaustive Item-by-Item Audit\n\nUnless the user explicitly requests sampling, you must check every reference one by one.\n\nYou must not:\n- Skip references because the list is long.\n- Only check suspicious-looking references.\n- Summarize without producing a per-reference audit table.\n- Mark a reference as verified based only on plausibility.\n- Collapse multiple references into one generic comment.\n- Stop without telling the user exactly which reference numbers have been checked and which remain unchecked.\n- Treat Chinese references as unverifiable merely because they lack DOI, PMID, or English metadata.\n- Require DOI for Chinese references unless the user, target journal, or citation style explicitly requires DOI.\n- Treat title-translation differences as errors unless the reference identity changes.\n\nIf the reference list is too long for one response, process it in sequential batches. Continue from the last checked reference number in the next round.\n\n## Required Data Fields for Each Reference\n\nFor every reference, extract and display the following fields whenever available:\n\n- Reference number\n- Original title\n- Translated title, if present\n- First author or submitted author string\n- Journal / book / conference / source\n- Year\n- Volume\n- Issue\n- Pages or article number\n- DOI, if provided or required by the target citation style\n- PMID / PMCID, if available\n- CNKI / Wanfang / VIP / official source traceability for Chinese references, if available or searched\n- ISBN, for books\n- Degree type and institution, for dissertations\n- Standard number, for standards\n- URL and access date, for webpages when relevant\n\nThe audit table must include enough bibliographic information for the user to identify the"},{"path":"README.md","content":"
\n\n# Reference Checker Skill｜参考文献核查 Skill\n\n🌐 Language: 中文 | [English](README.md)\n\n**不要让一条假文献毁掉整篇论文的可信度。**\n\nReference Checker Skill 是一个用于**学术论文投稿前参考文献逐条核查**的 AI Skill，适合研究生、科研作者、编辑、审稿助理和学术写作者使用。\n\n它的目标不是“帮你简单看一下参考文献”，而是让 AI 按照严格流程，对参考文献列表进行**逐条、结构化、可追踪的核验**，尽量发现投稿前最容易被忽略、但最可能影响论文可信度的问题。\n\nv1.4.0 起，本 Skill 明确支持**英文文献与中文文献的双语核查**。除 DOI、PubMed、Crossref、出版社页面等常规路径外，中文文献会优先使用**中文原题名**进行核查，并将**知网 CNKI、万方 Wanfang、维普 VIP/CQVIP、期刊官网、出版社目录、大学机构库、国家标准/政府官网**等纳入检索路径。\n\n---\n\n## v1.4.0 更新重点\n\n这一版主要增强了中文文献核查能力：\n\n* **支持中文文献逐条检查**：中文期刊论文、中文学位论文、中文图书、中文会议论文、标准、政策文件、专利等均可纳入核查。\n* **增加中文数据库检索路径**：中文期刊优先按“中文原题名”检索，推荐依次核对期刊官网、知网、万方、维普等来源。\n* **避免误判无 DOI 中文文献**：很多中文文献没有 DOI、PMID 或英文元数据，Skill 会要求 AI 不得仅因缺少这些字段就判定为不可验证。\n* **增加中文溯源字段**：输出表格中的 `DOI / PMID` 扩展为 `DOI / PMID / Chinese Trace`，便于记录 CNKI、万方、维普、期刊官网、ISBN、标准号、专利号等信息。\n* **增强中文格式检查**：新增对中文期刊名、中文作者名、GB/T 7714 风格、中文标点、页码、卷期、学位授予单位、出版社、ISBN 等字段的核查要求。\n* **区分中文原题名与英文译名**：对于双语中文文献，应优先核验中文题名；英文译名不一致时，只有在影响文献身份识别时才标记为严重问题。\n\n---\n\n## 为什么需要这个 Skill？\n\n在 AI 辅助写作越来越普遍的今天，参考文献错误变得更隐蔽，也更危险。\n\n很多错误看起来“很像真的”：\n\n* DOI 存在，但指向的是另一篇文章\n* 题名看起来合理，但数据库里查不到\n* 作者、期刊、年份、卷期页码彼此错配\n* 中文题名、英文译名、期刊英文名之间不一致\n* 中文文献没有 DOI，被错误地当作“无法核验”\n* 预印本已经有正式发表版本，但文中仍引用旧版本\n* 同一篇文献被重复引用\n* AI 生成了格式完整、内容逼真、但实际不存在的参考文献\n\n这些问题如果出现在投稿稿件中，轻则影响编辑和审稿人的第一印象，重则直接损害整篇论文的可信度。\n\n**Reference Checker Skill 的核心价值，就是在投稿前帮你把这些问题尽可能提前暴露出来。**\n\n---\n\n## 这个 Skill 能做什么？\n\nReference Checker Skill 会要求 AI 对参考文献进行**逐条核查**，而不是只挑几条看起来可疑的文献抽样检查。\n\n它会尽量核验每一条参考文献的关键信息，包括：\n\n| 检查项目 | 是否检查 |\n| --- | --- |\n| 文献是否真实存在 | 是 |\n| 文章题名 / 中文原题名 | 是 |\n| 作者信息 / 中文作者名 | 是 |\n| 期刊 / 来源 / 出版社 / 授予单位 | 是 |\n| 发表年份 / 出版年份 / 授予年份 | 是 |\n| 卷、期、页码或文章编号 | 是 |\n| DOI | 如有则检查 |\n| PMID / PMCID | 如有则检查 |\n| CNKI / 万方 / 维普 / 期刊官网溯源 | 中文文献优先检查 |\n| ISBN / 标准号 / 专利号 / 文件号 | 对应文献类型检查 |\n| 重复引用 | 是 |\n| 预印本与正式发表版本冲突 | 是 |\n| 撤稿或来源风险 | 可检测时检查 |\n| 中文/英文题名不一致 | 是 |\n| GB/T 7714 或目标期刊格式一致性 | 可按需求检查 |\n\n---\n\n## 核心特点\n\n### 1. 逐条核查，而不是抽样检查\n\n很多参考文献检查工具只适合快速浏览，容易漏掉隐藏错误。\n\n这个 Skill 默认要求 AI **逐条检查每一条参考文献**。除非用户明确要求抽样，否则不应跳过任何一条。\n\n---\n\n### 2. 不只查 DOI，还查元数据是否匹配\n\n一个 DOI 能打开，并不代表这条参考文献就是正确的。\n\n这个 Skill 会要求 AI 同时核对：\n\n* 题名是否一致\n* 作者是否匹配\n* 期刊名称是否正确\n* 年份、卷期、页码是否对应\n* DOI 是否真的指向该文章\n* PMID / PMCID 是否与文献信息一致\n* 中文文献是否能通过知网、万方、维普、期刊官网或其他可靠来源溯源\n\n它关注的不是“有没有 DOI”，而是**整条参考文献是否真实、准确、对应一致**。\n\n---\n\n### 3. 支持中文文献数据库检索\n\n对于中文文献，Skill 会要求 AI 优先按**中文原题名**核查，而不是直接用英文译名检索。\n\n推荐检索路径包括：\n\n| 文献类型 | 推荐核查路径 |\n| --- | --- |\n| 中文期刊论文 | 期刊官网、知网 CNKI、万方、维普、DOI 元数据 |\n| 中文学位论文 | 知网博硕士论文库、万方学位论文库、学校机构库、国家图书馆记录 |"},{"path":"_meta.json","content":"{\n \"ownerId\": \"kn72ayzheax209248fk1zdzfrx85h3ah\",\n \"slug\": \"reference-checker\",\n \"version\": \"1.0.2\",\n \"publishedAt\": 1780487800161\n}"},{"path":"skill-card.md","content":"## Description:\n\nExhaustively verify English and Chinese manuscript references before journal submission.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nResearchers, graduate students, editors, reviewers, and academic writers use this skill to audit manuscript reference lists item by item for authenticity, metadata accuracy, identifier traceability, source-type risks, duplication, and citation-format consistency.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Reference verification can require searching manuscript citation details against external scholarly or official sources.\n\nMitigation: Provide only the reference-list content needed for verification, and avoid sharing confidential manuscript content unless external lookup is acceptable.\n\nRisk: The skill may flag citation issues based on incomplete, paywalled, conflicting, or inaccessible metadata.\n\nMitigation: Use the audit table as review guidance and manually confirm ambiguous or manual-check items before journal submission.\n\n## Reference(s):\n\n- [Reference Checker Skill on ClawHub](https://clawhub.ai/liuxiangjian-ai/skills/reference-checker)\n- [Publisher profile: liuxiangjian-ai](https://clawhub.ai/user/liuxiangjian-ai)\n\n## Skill Output:\n\n**Output Type(s):** [Analysis, Markdown, Guidance]\n\n**Output Format:** [Markdown audit tables with concise summaries and follow-up guidance]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May be produced in sequential batches for long reference lists, with each round stating which references were checked and which remain.]\n\n## Skill Version(s):\n\n1.0.2 (source: server release metadata; artifact frontmatter lists 1.4.1)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":null,"editorialQuality":{"score":100,"threshold":65,"status":"thin","wordCount":1526,"uniquenessScore":44,"reasons":["uniqueness-below-45"]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":"No screenshots, media assets, or demo links are available."},"primaryImageUrl":null,"mediaAssetCount":0,"assets":[],"demoUrl":null},"ownerResources":{"evidence":{"source":"unclaimed","verified":false,"confidence":"low","updatedAt":"2026-10-11T06:35:13.296Z","emptyReason":"This page has not been claimed by the agent owner."},"hasCustomPage":false,"customPageUpdatedAt":null,"customLinks":[],"structuredLinks":{"docsUrl":null,"demoUrl":null,"supportUrl":null,"pricingUrl":null,"statusUrl":null},"customPage":null},"relatedAgents":{"evidence":{"source":"protocol-neighbors","verified":false,"confidence":"medium","updatedAt":"2026-10-11T08:42:24.654Z","emptyReason":null},"items":[{"id":"8ebccd8e-3863-4187-8355-c3f14e1f9edf","entityType":"agent","canonicalPath":"/agent/iofficeai-aionui","slug":"iofficeai-aionui","name":"AionUi","description":"Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!","url":"https://github.com/iOfficeAI/AionUi","homepage":"https://www.aionui.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-10-09T19:11:12.944Z","createdAt":"2026-02-25T03:38:16.584Z","downloads":null},{"id":"b917f68a-ebff-438e-84f8-3f4b2494c0bc","entityType":"agent","canonicalPath":"/agent/activepieces-activepieces","slug":"activepieces-activepieces","name":"activepieces","description":"AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents","url":"https://github.com/activepieces/activepieces","homepage":"https://www.activepieces.com","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-15T02:22:12.426Z","createdAt":"2026-02-25T03:38:12.412Z","downloads":null},{"id":"5cb26759-3a39-483f-94cf-276a98c13bb8","entityType":"agent","canonicalPath":"/agent/cherryhq-cherry-studio","slug":"cherryhq-cherry-studio","name":"cherry-studio","description":"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs","url":"https://github.com/CherryHQ/cherry-studio","homepage":"https://cherry-ai.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-11T14:38:40.986Z","createdAt":"2026-02-25T03:38:19.379Z","downloads":null},{"id":"6f6582d0-5d76-4f0f-b81d-86520247950b","entityType":"agent","canonicalPath":"/agent/copilotkit-copilotkit","slug":"copilotkit-copilotkit","name":"CopilotKit","description":"The Frontend for Agents & Generative UI. React + Angular","url":"https://github.com/CopilotKit/CopilotKit","homepage":"https://docs.copilotkit.ai","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-03-25T09:50:57.846Z","createdAt":"2026-02-25T03:39:14.617Z","downloads":null}],"links":{"hub":"/agent","source":"/agent/source/clawhub","protocols":[{"label":"OpenClaw","href":"/agent/protocol/openclew"}]}}}
![\"中文\"](\"https://img.shields.io/badge/语言-中文-green\"){kind=icon}
\n \n ![\"English\"](\"https://img.shields.io/badge/Language-English-blue\"){kind=icon}
\n \n ![\"AI](\"https://img.shields.io/badge/类型-AI%20Skill-purple\"){kind=icon}
\n ![\"DOI](\"https://img.shields.io/badge/核查-DOI%20%2B%20元数据-orange\"){kind=icon}
\n ![\"中文文献核查\"](\"https://img.shields.io/badge/中文文献-CNKI%20%2B%20万方%20%2B%20维普-red\"){kind=icon}
\n ![\"v1.4.0\"](\"https://img.shields.io/badge/版本-v1.4.0-lightgrey\"){kind=icon}
\n ![\"MIT](\"https://img.shields.io/badge/许可证-MIT-yellow\"){kind=icon}
\n