{"id":"07d0edbd-9f21-482c-9e94-c2dc1a283ec6","entityType":"agent","slug":"clawhub-tianzhiceng297-boop-patent-gap-supply-chain","name":"patent-gap-supply-chain","canonicalUrl":"https://www.xpersona.co/agent/clawhub-tianzhiceng297-boop-patent-gap-supply-chain","canonicalPath":"/agent/clawhub-tianzhiceng297-boop-patent-gap-supply-chain","generatedAt":"2026-10-10T23:50:03.588Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T21:09:34.170Z","emptyReason":null},"description":"Competitive intelligence workflow that infers supply-chain relationships from patent gaps — with trade secret assessment, competitive triangulation, negative verification, patent quality scoring, evidence freshness tracking, reverse workflow (supplier→client), and financial cross-validation. 通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、 负向验证、专利质量评分、证据时效追踪、反向工作流（供应商→客户）、财务数据交叉验证于一体。 Use when researching tech/hardware/manufacturing/pharma stocks, analyzing industry supply chains, identifying suppliers or customers, investigating patent gaps, or inferring supply-chain relationships from patent ownership. Supports both forward (patent gap → supplier) and reverse (technology owner → downstream clients) analysis directions. 触发场景：科技股研究、行业分析、产业链推断、 供应商识别、客户挖掘、专利缺口分析、专利归属查询、技术依赖分析、上下游推断 / Triggers: patent gap analysis, supply chain inference, supplier identification, customer identification, technology dependency, competitive intelligence, tech stock research, industry analysis, trade secret assessment, negative verification, patent quality, financial cross-validation, reverse supply-chain analysis. The skill starts from a patent gap (or known technology owner), evaluates trade secret likelihood, activates five verification leads with competitive triangulation, runs negative verification to rule out false positives, scores patent quality, tracks evidence freshness, then cross-validates against financial disclosures, outputting a confidence-scored supply-chain inference report.","descriptionLabel":"Source description","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1.3K downloads reported by the source. Last updated 10/10/2026.","installCommand":"clawhub skill install s1708rbjpe5b8xjv91f2whm2k9883dny:patent-gap-supply-chain","sourceUrl":"https://clawhub.ai/tianzhiceng297-boop/patent-gap-supply-chain","homepage":"https://clawhub.ai/tianzhiceng297-boop/skills/patent-gap-supply-chain","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/tianzhiceng297-boop/patent-gap-supply-chain","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/tianzhiceng297-boop/skills/patent-gap-supply-chain","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":62,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"patent-gap-supply-chain technical dossier on Xpersona with agent coverage, OPENCLEW support, and live trust metadata."},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-10T21:09:34.170Z","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":"2026-10-10T21:09:34.170Z","emptyReason":null},"stars":null,"forks":null,"downloads":1257,"packageName":null,"latestVersion":"0.1.2","tractionLabel":"1.3K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-10T21:09:34.164Z","emptyReason":null},"lastUpdatedAt":"2026-10-10T21:09:34.170Z","lastCrawledAt":"2026-10-10T21:09:34.164Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-11T21:09:34.164Z","lastVerifiedAt":null,"highlights":[{"version":"0.1.2","createdAt":"2026-08-06T02:50:16.328Z","changelog":"v0.4.0: patent quality scoring + negative verification + evidence freshness tracking","fileCount":10,"zipByteSize":39280},{"version":"0.1.1","createdAt":"2026-07-21T05:44:02.274Z","changelog":"**v0.3.0 is a major upgrade with new reverse-analysis and validation features:** - Added reverse-supply-chain workflow: supports starting from a technology owner to find downstream clients, not just supplier identification from a patent gap. - Introduced competitive triangulation: boosts inference confidence if multiple competitors show the same patent gap pointing to the same supplier. - Integrated financial cross-validation: cross-checks findings against disclosed supplier/customer lists and financials for higher confidence. - Built-in trade secret assessment: explicitly distinguishes between genuine patent gaps and cases where technology is kept as a trade secret. - Expanded resources and guidance: added industry-specific workflow tips and detailed examples for various verticals. - Updated documentation and examples to reflect all new features and analysis directions.","fileCount":11,"zipByteSize":33310},{"version":"0.1.0","createdAt":"2026-07-09T04:20:22.876Z","changelog":"- Initial release of the \"patent-gap-supply-chain\" skill. - Implements a 7-step competitive intelligence workflow to infer supply-chain relationships from patent gaps. - Activates five verification leads in parallel: patent ownership, corporate affiliation, talent flow, procurement, and product teardown. - Includes mandatory alternative-hypothesis and counter-intelligence checks to prevent overconfident or incorrect conclusions. - Outputs a structured report with relationship determination, evidence chain, confidence scoring, and follow-up suggestions. - Provides clear applicability guidance with pre-check for supported target types and alternative advice for unsupported cases.","fileCount":3,"zipByteSize":4539}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s1708rbjpe5b8xjv91f2whm2k9883dny:patent-gap-supply-chain","setupComplexity":"low","setupSteps":["Install using `clawhub skill install s1708rbjpe5b8xjv91f2whm2k9883dny:patent-gap-supply-chain` 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/tianzhiceng297-boop/patent-gap-supply-chain 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-tianzhiceng297-boop-patent-gap-supply-chain/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-10T23:50:03.584Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-tianzhiceng297-boop-patent-gap-supply-chain/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":"2026-10-10T21:09:34.170Z","emptyReason":null},"readme":"Skill: patent-gap-supply-chain\n\nOwner: tianzhiceng297-boop\n\nSummary: Competitive intelligence workflow that infers supply-chain relationships from patent gaps — with trade secret assessment, competitive triangulation, negative verification, patent quality scoring, evidence freshness tracking, reverse workflow (supplier→client), and financial cross-validation. 通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、 负向验证、专利质量评分、证据时效追踪、反向工作流（供应商→客户）、财务数据交叉验证于一体。 Use when researching tech/hardware/manufacturing/pharma stocks, analyzing industry supply chains, identifying suppliers or customers, investigating patent gaps, or inferring supply-chain relationships from patent ownership. Supports both forward (patent gap → supplier) and reverse (technology owner → downstream clients) analysis directions. 触发场景：科技股研究、行业分析、产业链推断、 供应商识别、客户挖掘、专利缺口分析、专利归属查询、技术依赖分析、上下游推断 / Triggers: patent gap analysis, supply chain inference, supplier identification, customer identification, technology dependency, competitive intelligence, tech stock research, industry analysis, trade secret assessment, negative verification, patent quality, financial cross-validation, reverse supply-chain analysis. The skill starts from a patent gap (or known technology owner), evaluates trade secret likelihood, activates five verification leads with competitive triangulation, runs negative verification to rule out false positives, scores patent quality, tracks evidence freshness, then cross-validates against financial disclosures, outputting a confidence-scored supply-chain inference report.\n\nTags: latest:0.1.2\n\nVersion history:\n\nv0.1.2 | 2026-08-06T02:50:16.328Z | user\n\nv0.4.0: patent quality scoring + negative verification + evidence freshness tracking\n\nv0.1.1 | 2026-07-21T05:44:02.274Z | auto\n\n**v0.3.0 is a major upgrade with new reverse-analysis and validation features:**\n\n- Added reverse-supply-chain workflow: supports starting from a technology owner to find downstream clients, not just supplier identification from a patent gap.\n- Introduced competitive triangulation: boosts inference confidence if multiple competitors show the same patent gap pointing to the same supplier.\n- Integrated financial cross-validation: cross-checks findings against disclosed supplier/customer lists and financials for higher confidence.\n- Built-in trade secret assessment: explicitly distinguishes between genuine patent gaps and cases where technology is kept as a trade secret.\n- Expanded resources and guidance: added industry-specific workflow tips and detailed examples for various verticals.\n- Updated documentation and examples to reflect all new features and analysis directions.\n\nv0.1.0 | 2026-07-09T04:20:22.876Z | auto\n\n- Initial release of the \"patent-gap-supply-chain\" skill.\n- Implements a 7-step competitive intelligence workflow to infer supply-chain relationships from patent gaps.\n- Activates five verification leads in parallel: patent ownership, corporate affiliation, talent flow, procurement, and product teardown.\n- Includes mandatory alternative-hypothesis and counter-intelligence checks to prevent overconfident or incorrect conclusions.\n- Outputs a structured report with relationship determination, evidence chain, confidence scoring, and follow-up suggestions.\n- Provides clear applicability guidance with pre-check for supported target types and alternative advice for unsupported cases.\n\nArchive index:\n\nArchive v0.1.2: 10 files, 39280 bytes\n\nFiles: CHANGELOG.md (7432b), examples/real-case.md (9886b), examples/sample-output.md (10132b), README.md (6954b), references/checklists.md (14772b), references/industry-guide.md (9161b), references/workflow.md (17208b), skill-card.md (2880b), SKILL.md (14396b), _meta.json (142b)\n\nFile v0.1.2:SKILL.md\n\n---\r\nname: patent-gap-supply-chain\r\ndescription: 'Competitive intelligence workflow that infers supply-chain\r\n  relationships from patent gaps — with trade secret assessment, competitive\r\n  triangulation, negative verification, patent quality scoring, evidence\r\n  freshness tracking, reverse workflow (supplier→client), and financial\r\n  cross-validation. 通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、\r\n  负向验证、专利质量评分、证据时效追踪、反向工作流（供应商→客户）、财务数据交叉验证于一体。\r\n  Use when researching tech/hardware/manufacturing/pharma stocks, analyzing\r\n  industry supply chains, identifying suppliers or customers, investigating\r\n  patent gaps, or inferring supply-chain relationships from patent ownership.\r\n  Supports both forward (patent gap → supplier) and reverse (technology owner →\r\n  downstream clients) analysis directions. 触发场景：科技股研究、行业分析、产业链推断、\r\n  供应商识别、客户挖掘、专利缺口分析、专利归属查询、技术依赖分析、上下游推断 / Triggers:\r\n  patent gap analysis, supply chain inference, supplier identification,\r\n  customer identification, technology dependency, competitive intelligence,\r\n  tech stock research, industry analysis, trade secret assessment, negative\r\n  verification, patent quality, financial cross-validation, reverse\r\n  supply-chain analysis. The skill starts from a patent gap (or known technology\r\n  owner), evaluates trade secret likelihood, activates five verification leads\r\n  with competitive triangulation, runs negative verification to rule out false\r\n  positives, scores patent quality, tracks evidence freshness, then\r\n  cross-validates against financial disclosures, outputting a\r\n  confidence-scored supply-chain inference report.'\r\nversion: 0.4.0\r\nagent_created: true\r\ndisable: true\r\n---\r\n\r\n# Patent Gap Supply Chain Inference\r\n\r\n## Overview\r\n\r\nThis skill implements a competitive intelligence workflow: \"patent gap → supply chain inference.\" Core insight: when target company A's product uses technology T, but A holds no patents related to T, there may be a technology-licensing or supply-chain dependency. By searching T's patent ownership to identify company B, then activating multiple verification leads, one can infer whether B has entered A's supply chain.\r\n\r\n**Critical caveat**: A patent gap can also mean A protects T as a **trade secret / know-how** rather than through patents. Patents require public disclosure; many companies (especially in semiconductor processes, chemical formulations, pharma manufacturing) deliberately choose trade secrets. This skill explicitly evaluates this distinction before concluding supply-chain dependency.\r\n\r\n**v0.4.0 new capabilities**:\r\n- **Patent quality scoring**: Not all patents are equal. A core patent (broad claims, high citations, PCT family) carries far more weight than a peripheral patent (narrow claims, single jurisdiction). Lead 1 now scores patent quality, not just quantity.\r\n- **Negative verification**: Actively search for evidence that B is NOT A's supplier (revenue mismatch, capacity mismatch, geography, explicit customer lists excluding A). Only when negative evidence fails does the positive chain stand. This is the strongest defense against false positives.\r\n- **Evidence freshness tracking**: Every lead carries an evidence timestamp. Evidence older than 18 months is flagged and down-weighted; conclusions resting on stale evidence cannot reach \"high\" confidence.\r\n- **Quality × triangulation interplay**: Patent quality scoring feeds into triangulation — peers converging on the same HIGH-QUALITY patent holder is much stronger than convergence on a peripheral patent holder.\r\n\r\nThe skill downgrades \"patents\" from the sole evidence chain to an entry lead, enforcing multi-lead parallel verification, competitive triangulation, negative verification, financial cross-validation, alternative-hypothesis checks, and counter-intelligence checks to prevent overconfident wrong conclusions.\r\n\r\n## When to Use\r\n\r\nTrigger when any of the following applies:\r\n\r\n**Forward direction** (patent gap → find supplier):\r\n- Researching tech, hardware, manufacturing, pharma, or materials companies where supply chain matters\r\n- Discovering or suspecting a company uses a technology but lacks related patents\r\n- Explicitly asked to do patent ownership analysis, technology dependency analysis, or supply-chain inference\r\n\r\n**Reverse direction** (technology owner → find clients):\r\n- A company B holds dominant/monopoly patents in a technology T and you want to identify B's downstream customers\r\n- Investment thesis: \"B is the sole/dominant supplier of T; companies using T must be B's customers\"\r\n- Asked to identify companies that depend on a specific technology or component\r\n\r\nTrigger keywords (bilingual): 科技股研究, 行业分析, 产业链, 供应链, 供应商识别, 客户挖掘, 上下游推断, 专利缺口, 专利归属, 技术依赖, 专利质量, 负向验证, patent gap, supply chain, supplier identification, customer identification, technology dependency, competitive intelligence, patent quality, negative verification, reverse supply chain.\r\n\r\n## Step 0: Applicability Pre-check (mandatory)\r\n\r\nBefore running the workflow, judge whether the target type is applicable. See `references/industry-guide.md` for industry-specific dynamics:\r\n\r\n| Target type | Applicability | Handling |\r\n|-------------|---------------|----------|\r\n| Hardware / manufacturing / pharma / materials | Strong | Proceed. Reference `industry-guide.md` for vertical-specific signals |\r\n| Software / internet | Weak | Warn that patent protection is weak and tech is hidden; suggest code / open-source intelligence; may proceed at reduced confidence |\r\n| Business-model innovation | Not applicable | Stop; suggest financial reports / prospectus / business-model analysis instead |\r\n\r\nWhen not applicable, clearly inform the user and suggest alternatives. Do not force execution.\r\n\r\n## Direction Selection\r\n\r\nBefore proceeding, determine the analysis direction:\r\n\r\n| Direction | Starting point | Goal | Trigger phrase examples |\r\n|-----------|---------------|------|------------------------|\r\n| **Forward** | Company A uses T but lacks T's patents | Find who supplies T to A | \"Identify XX's suppliers\", \"Who provides XX's technology?\" |\r\n| **Reverse** | Company B holds dominant T patents | Find who depends on B's T | \"Who are B's customers?\", \"What downstream companies need this patent?\" |\r\n\r\nBoth directions share the same verification leads, but the entry point and initial search differ. See `references/workflow.md` for detailed reverse-workflow operations.\r\n\r\n## Workflow Overview\r\n\r\nThe full workflow has 11 steps. Detailed operations in `references/workflow.md`:\r\n\r\n1. **Trigger lead identification**: Confirm target company, product, technology; determine forward or reverse direction\r\n2. **Patent gap confirmation** (forward) / **Patent dominance confirmation** (reverse): Search patent portfolio, determine gap validity or market dominance\r\n3. **Trade secret assessment**: Evaluate whether gap reflects deliberate trade secret / know-how protection\r\n4. **Multi-lead parallel verification**: Activate 5 leads — patent ownership, corporate affiliation, talent flow, procurement, product teardown. **Patent quality scoring** applied to Lead 1.\r\n5. **Competitive triangulation**: Check A's peers/competitors for the same patent gap. If ≥2 peers also lack T's patents and all point to B, boost confidence.\r\n6. **Negative verification** (NEW v0.4.0): Actively search for evidence that B is NOT A's supplier. Only when negative evidence fails does the positive chain stand.\r\n7. **Multi-lead fusion + confidence scoring**: ≥2 high-confidence leads corroborating required; triangulation provides confidence multiplier; **evidence freshness adjustment** applied (NEW v0.4.0).\r\n8. **Financial cross-validation**: Cross-validate inferred relationship against A's supplier disclosures / B's customer concentration / accounts receivable. Inference without financial corroboration is capped.\r\n9. **Alternative-hypothesis check**: Explicitly exclude alternative explanations; see `references/checklists.md`\r\n10. **Counter-intelligence check**: Check for shell companies / dispersed patents; see `references/checklists.md`\r\n11. **Output report**: Relationship determination + evidence chain + patent quality + evidence freshness + negative verification + triangulation + financial corroboration + confidence + follow-up\r\n\r\n## Input/Output Contract\r\n\r\n**Input**:\r\n- Target company (A for forward, B for reverse) — name / ticker\r\n- Technology T or product P of interest (optional; may be identified during analysis)\r\n- Analysis direction: forward or reverse (default: forward if user mentions \"supplier\", reverse if user mentions \"customer\" or \"downstream\")\r\n- Known background info (optional)\r\n\r\n**Output**: Supply-chain inference report, including:\r\n- **Direction**: forward or reverse analysis\r\n- **Relationship determination**: whether a supply-chain relationship exists + role (direct supplier / technology licensor / customer / indirect affiliation / NPE license-only)\r\n- **Trade secret assessment**: likelihood + indicator breakdown\r\n- **Evidence chain**: per-lead hit status + per-lead confidence + **patent quality score** (for Lead 1) + **evidence timestamp/freshness** (all leads, v0.4.0)\r\n- **Triangulation result**: which peers confirmed the same gap + triangulation boost applied\r\n- **Negative verification result** (v0.4.0): negative evidence searched + found/not found + impact on confidence\r\n- **Financial corroboration**: supplier disclosure hits, customer concentration data, AR patterns, corroboration status\r\n- **Overall confidence**: high / medium / low + rationale (including all adjustments: trade secret, triangulation, negative verification, evidence freshness, financial cap)\r\n- **Alternative-hypothesis list**: excluded + unexcluded\r\n- **Counter-intelligence risk flag**: shell companies / dispersed patents\r\n- **Follow-up verification suggestions**: next evidence to gather\r\n\r\n## Tool Integration\r\n\r\nAvailable tools per lead (by availability priority):\r\n\r\n| Lead | Tools | Purpose |\r\n|------|-------|---------|\r\n| Patent search + quality | tushare (patent API), PatSnap patsnap-search connector | Search A/B/peers patents, T's patent ownership, patent quality scoring (claims, citations, family) |\r\n| Corporate affiliation | Tianyancha tyc-mcp, Qichacha qcc-company, Qixinbao qixinbao-mcp | A-B investment / JV / executive overlap |\r\n| Listed-co data | tushare, westock-mcp, westock-tool | Financial report supplier disclosure, prospectus |\r\n| Talent flow | WebSearch, job-site search | B→A technical hires |\r\n| Procurement | WebSearch, gov-procurement portals, tender announcements | A's tender / supplier announcements |\r\n| Product teardown | WebSearch, professional teardown reports | A's product BOM / teardown for B components |\r\n| Trade secret assessment | WebSearch, legal databases, company reports | NDAs, non-competes, trade secret litigation, R&D disclosures |\r\n| Competitive triangulation | tushare, WebSearch, industry reports | A's peer patent portfolios, cross-company gap confirmation |\r\n| Negative verification (v0.4.0) | WebSearch, tushare, annual reports, B's customer disclosures | Revenue mismatch, capacity mismatch, geography, explicit customer lists |\r\n| Financial cross-validation | westock-mcp, tushare, annual report filings | Supplier concentration, customer concentration, AR aging, related-party transactions |\r\n| Evidence freshness (v0.4.0) | All sources | Timestamp every evidence item; flag >18 months old |\r\n| Industry guidance | `references/industry-guide.md` | Vertical-specific patent/trade-secret dynamics and verification strategies |\r\n\r\nWhen a tool is unavailable, fall back to WebSearch for public info and reduce that lead's confidence accordingly.\r\n\r\n## Key Principles\r\n\r\n- **Patents are an entry lead, not final evidence**: A patent gap triggers verification but does not directly conclude\r\n- **Trade secrets must be ruled out before concluding dependency**: A patent gap may reflect deliberate trade secret / know-how protection\r\n- **Procurement/teardown leads override trade secret hypothesis**: Physical evidence trumps trade secret likelihood\r\n- **Patent quality matters more than patent count** (v0.4.0): A core patent family with broad claims outweighs dozens of peripheral patents. Score quality, not just presence.\r\n- **Negative verification is mandatory** (v0.4.0): You must actively attempt to disprove the inference. An inference that survives active disproof attempts is far more credible than one that was never tested.\r\n- **Evidence freshness is tracked** (v0.4.0): Every lead carries an evidence timestamp. Stale evidence (>18 months) cannot support \"high\" confidence alone.\r\n- **Competitive triangulation strengthens inference**: Cross-company corroboration is one of the strongest signals\r\n- **Financial cross-validation is required for high-confidence conclusions**: No inference reaches \"high\" without financial corroboration\r\n- **Multi-lead corroboration**: ≥2 high-confidence leads corroborating required to conclude\r\n- **Explicit confidence scoring**: Every lead and inference gets a confidence score (high / medium / low)\r\n- **Alternative hypotheses are mandatory**: Every inference must explicitly exclude alternative explanations\r\n- **Counter-intelligence check is mandatory**: Watch for shell companies / dispersed patents hiding the real supply chain\r\n\r\n## Resources\r\n\r\n### references/\r\n- `references/workflow.md`: Detailed 11-step operations + reverse workflow section\r\n- `references/checklists.md`: All checklists — alternative hypotheses, trade secret indicators, patent quality scoring, negative verification, evidence freshness, triangulation, financial cross-validation, counter-intelligence, confidence scoring rules\r\n- `references/industry-guide.md`: Vertical-specific guidance for semiconductor, pharma, new energy, consumer electronics, and materials\r\n\r\n### examples/\r\n- `examples/sample-output.md`: Fictional case showing the full 11-step report format\r\n- `examples/real-case.md`: A real Chinese A-share case study with actual company names and data\n\nFile v0.1.2:README.md\n\n# patent-gap-supply-chain\r\n\r\n通过\"专利缺口\"反推产业链关系的竞争情报工作流 skill，适用于 WorkBuddy / Codebuddy / Claude Code / Codex。\r\n\r\nA competitive intelligence workflow skill that infers supply-chain relationships from patent gaps. Works with WorkBuddy / Codebuddy / Claude Code / Codex.\r\n\r\n---\r\n\r\n## 中文说明\r\n\r\n### 这是什么\r\n\r\n当研究科技股 / 硬件 / 制造 / 医药类公司时，发现标的 A 的产品使用了技术 T，但 A 没有 T 的相关专利——这套 skill 引导从专利缺口切入，**先判断缺口是真实外部依赖还是商业秘密/know-how保护**，再激活五条线索并行验证，竞对三角验证增强信号，**负向验证排除假阳性**，**专利质量评分识别真技术持有者**，**证据时效追踪避免过时结论**，财务数据交叉验证定论，输出带置信度的产业链关系推断报告。\r\n\r\n**v0.4.0 新增**：\r\n- **专利质量评分**：不看数量看质量——权利要求宽度、被引影响、同族规模、法律状态四维评分，核心专利(≥8/12)才支撑高置信度\r\n- **负向验证**：主动寻找\"B 不是 A 供应商\"的证据（收入错配/产品错配/地理不可行/客户名单排除），反证失败正向证据链才成立\r\n- **证据时效追踪**：每条证据带时间戳，>18个月标记为过时并降权，结论不能只靠过时证据\r\n\r\n**v0.3.0**：竞对三角验证 + 财务交叉验证 + 反向工作流 + 行业指南 + 真实案例\r\n**v0.2.0**：商业秘密/know-how 评估\r\n\r\n### 核心设计\r\n\r\n- **商业秘密先排除再定论**：专利缺口可能是商业秘密保护策略，必须先评估\r\n- **专利质量 > 专利数量** (v0.4.0)：一份核心专利胜过几十份外围专利\r\n- **负向验证必做** (v0.4.0)：主动尝试推翻推断，经受住反证的推断才可信\r\n- **证据时效必标** (v0.4.0)：过时证据不能单独支撑高置信度\r\n- **竞对三角验证**：多同行同一缺口 → 同一供应商，是最强信号之一\r\n- **财务交叉验证**：无财报佐证的推断置信度最多\"中等\"\r\n- **双向工作流**：正向（缺口→供应商）+ 反向（专利所有者→下游客户）\r\n- **专利降级为入口线索**：专利缺口只触发验证，不直接定论\r\n\r\n### 安装\r\n\r\n    git clone https://github.com/tianzhiceng297-boop/patent-gap-supply-chain.git ~/.workbuddy/skills/patent-gap-supply-chain\r\n\r\n或手动复制 `SKILL.md` 和 `references/` 到 `~/.workbuddy/skills/patent-gap-supply-chain/`。\r\n\r\n### 触发场景\r\n\r\n- \"研究 XX 科技股\"\r\n- \"分析 XX 公司的供应链 / 供应商\"\r\n- \"查一下 XX 技术的专利归属\"\r\n- \"XX 公司用了某技术但好像没专利\"\r\n- \"识别 XX 的供应商\"\r\n- \"XX 公司的下游客户有哪些？\" (反向工作流)\r\n- \"哪些公司依赖 XX 的技术？\" (反向工作流)\r\n\r\n### 适用边界\r\n\r\n| 标的类型 | 适用性 |\r\n|---------|--------|\r\n| 硬件 / 制造 / 医药 / 材料 | 强适用 |\r\n| 软件 / 互联网 | 弱适用（降低置信度） |\r\n| 商业模式创新 | 不适用 |\r\n\r\n---\r\n\r\n## English\r\n\r\n### What it is\r\n\r\nWhen researching tech / hardware / manufacturing / pharma companies, you find that target A's product uses technology T, but A holds no patents related to T. This skill starts from the patent gap, **first evaluates whether the gap reflects true external dependency or deliberate trade secret / know-how protection**, then activates five parallel verification leads, competitive triangulation, **negative verification**, **patent quality scoring**, **evidence freshness tracking**, and financial cross-validation, outputting a confidence-scored supply-chain inference report.\r\n\r\n**New in v0.4.0**:\r\n- **Patent quality scoring**: 4-factor quality score (claim breadth, citations, family size, legal status). Core patents (≥8/12) support high confidence; peripheral patents cap at low.\r\n- **Negative verification**: Actively search for evidence that B is NOT A's supplier. Inference survives active disproof attempts or is down-weighted.\r\n- **Evidence freshness tracking**: Every evidence item carries a timestamp. Stale (>18 months) evidence is down-weighted; conclusions cannot rest primarily on stale evidence.\r\n\r\n**v0.3.0**: competitive triangulation + financial cross-validation + reverse workflow + industry guide + real case\r\n**v0.2.0**: trade secret / know-how assessment\r\n\r\n### Core design\r\n\r\n- **Trade secrets ruled out before concluding**: patent gap may reflect trade secret strategy\r\n- **Patent quality > patent count** (v0.4.0): one core patent outweighs dozens of peripheral ones\r\n- **Negative verification mandatory** (v0.4.0): actively try to disprove; surviving inferences are credible\r\n- **Evidence freshness tracked** (v0.4.0): stale evidence cannot support \"high\" alone\r\n- **Competitive triangulation**: multi-peer same gap → same supplier is a top-tier signal\r\n- **Financial cross-validation required**: no \"high\" without financial corroboration\r\n- **Bidirectional workflow**: forward (gap → supplier) + reverse (patent owner → downstream clients)\r\n- **Patents demoted to an entry lead**: a patent gap triggers verification but does not directly conclude\r\n\r\n### Installation\r\n\r\n    git clone https://github.com/tianzhiceng297-boop/patent-gap-supply-chain.git ~/.workbuddy/skills/patent-gap-supply-chain\r\n\r\nOr manually copy `SKILL.md` and `references/` to `~/.workbuddy/skills/patent-gap-supply-chain/`.\r\n\r\n### Trigger scenarios\r\n\r\n- \"Research [tech stock]\"\r\n- \"Analyze [company]'s supply chain / suppliers\"\r\n- \"Look up patent ownership of [technology]\"\r\n- \"[Company] uses a technology but seems to have no patents\"\r\n- \"Identify [company]'s suppliers\"\r\n- \"Who are [company]'s downstream customers?\" (reverse workflow)\r\n- \"Which companies depend on [technology]?\" (reverse workflow)\r\n\r\n### Applicability\r\n\r\n| Target type | Applicability |\r\n|-------------|---------------|\r\n| Hardware / manufacturing / pharma / materials | Strong |\r\n| Software / internet | Weak (reduce confidence) |\r\n| Business-model innovation | Not applicable |\r\n\r\n---\r\n\r\n## File structure\r\n\r\n- `SKILL.md` — core: triggers, direction selection, 11-step workflow (triangulation + negative verification + patent quality + evidence freshness + financial cross-validation), I/O contract, tool integration, key principles\r\n- `references/workflow.md` — detailed 11-step operations + reverse workflow appendix\r\n- `references/checklists.md` — all checklists including patent quality scoring, negative verification, evidence freshness, 5-layer confidence adjustment\r\n- `references/industry-guide.md` — vertical-specific guidance: semiconductor, pharma, new energy, consumer electronics, materials/chemicals\r\n- `examples/sample-output.md` — fictional case showing full 11-step report format\r\n- `examples/real-case.md` — real A-share case: Goertek & Apple VR supply chain\r\n- `CHANGELOG.md` — version history\r\n\r\n## License\r\n\r\nMIT\n\nFile v0.1.2:_meta.json\n\n{\n  \"ownerId\": \"kn718c4g4psbwaq7bt8ft547cd8827z6\",\n  \"slug\": \"patent-gap-supply-chain\",\n  \"version\": \"0.1.2\",\n  \"publishedAt\": 1785984616328\n}\n\nFile v0.1.2:references/checklists.md\n\n# Checklists and Confidence Rules\r\n\r\n## Alternative-Hypothesis Checklist\r\n\r\nFor each inference, check each item and explicitly exclude alternative explanations.\r\n\r\n### Alternative explanations for the patent gap\r\n- [ ] Does A's patent gap stem from a cross-licensing agreement? (check announcements / news)\r\n- [ ] Did A inherit T usage rights via an acquisition? (check M&A history)\r\n- [ ] Does T belong to a patent pool? (e.g., Avanci, MPEG-LA, Via Licensing)\r\n- [ ] Is there an open-source implementation of T? (check open-source projects)\r\n- [ ] Does A hold T patents via a subsidiary / affiliate? (check related-entity patents)\r\n- [ ] Does A deliberately protect T as a trade secret / know-how rather than patenting it? (run Trade Secret Indicators checklist)\r\n- [ ] Is the patent gap simply because T is a purchased component that A has no reason to patent? (common in consumer electronics — weak signal)\r\n\r\n### Alternative explanations for \"B entered the supply chain\"\r\n- [ ] Is B an NPE (non-practicing entity)? If so, B licenses only and is not in the supply chain\r\n- [ ] Does A use B's patent indirectly via a third-party module? The third party is the supplier; B is its upstream\r\n- [ ] Is B only a patent licensor with no physical transaction with A?\r\n- [ ] Is the A-B tie purely financial investment, not business cooperation?\r\n- [ ] Is the A-B relationship economically immaterial? (financial cross-validation required to exclude)\r\n\r\n## Patent Quality Scoring (NEW v0.4.0)\r\n\r\n**Principle**: Patent count is a weak signal. Quality matters. A core patent family outweighs dozens of peripheral patents. Apply this scoring to Lead 1 (patent ownership) and to Step 2B (dominance confirmation).\r\n\r\n### Factor 1 · Claim breadth (0-3)\r\n- [ ] 3: Independent claims covering the core mechanism/architecture, broad scope\r\n- [ ] 2: Claims cover a specific implementation, moderate scope\r\n- [ ] 1: Claims cover a narrow detail/variant\r\n- [ ] 0: Claims are process-only or trivial\r\n\r\n### Factor 2 · Citation impact (0-3)\r\n- [ ] 3: High forward citations (top decile in field); later patents build on it\r\n- [ ] 2: Moderate citations; some later work references it\r\n- [ ] 1: Few citations\r\n- [ ] 0: Never cited\r\n\r\n### Factor 3 · Family size (0-3)\r\n- [ ] 3: PCT + multiple national entries (US/EU/CN/JP/KR)\r\n- [ ] 2: PCT application, few national entries\r\n- [ ] 1: Single jurisdiction (e.g., CN only)\r\n- [ ] 0: Provisional/unpublished\r\n\r\n### Factor 4 · Legal status (0-3)\r\n- [ ] 3: Granted AND maintenance fees paid (active)\r\n- [ ] 2: Granted but recently filed; maintenance status unknown\r\n- [ ] 1: Under examination / pending\r\n- [ ] 0: Lapsed, revoked, abandoned, or transferred to shell\r\n\r\n### Quality levels\r\n- **Core patent**: total ≥8/12 — broad claims, high citations, international family, active legal status\r\n- **Supporting patent**: total 4-7/12\r\n- **Peripheral patent**: total 0-3/12 — narrow, uncited, single-jurisdiction, or lapsed\r\n\r\n### Impact on confidence\r\n- B holds Core patents in T → Lead 1 can reach \"high\"; triangulation convergence on B is stronger\r\n- B holds only Peripheral patents → Lead 1 capped at \"low\"; triangulation convergence is weaker (B may be a paper tiger)\r\n- A (forward) holds peripheral T patents → gap may be understated; note in report\r\n- If patent quality cannot be assessed (data unavailable) → treat as Supporting, flag in report\r\n\r\n## Negative Verification Checklist (NEW v0.4.0)\r\n\r\n**Principle**: Before concluding B is A's supplier, actively attempt to disprove it. Document the search.\r\n\r\n### Revenue / scale mismatch\r\n- [ ] Checked B's total revenue vs. A's plausible procurement of T\r\n- [ ] Checked if B's revenue is implausibly large/small relative to A's needs\r\n- [ ] Checked B's capacity expansion announcements for product alignment with A\r\n\r\n### Product-category mismatch\r\n- [ ] Checked B's actual product catalog for the specific component A uses\r\n- [ ] Checked if B's patents cover a different product variant than A uses\r\n- [ ] Checked whether a third party (C) manufactures the physical product under B's license\r\n\r\n### Geographic / logistics infeasibility\r\n- [ ] Checked geographic compatibility (regulatory, shipping cost, local-content requirements)\r\n- [ ] Checked trade restrictions / tariffs affecting A-B trade\r\n\r\n### Explicit customer-list exclusion\r\n- [ ] Checked if B publicly discloses customers (annual report, case studies, press releases)\r\n- [ ] If B's customer list is available: is A absent from it despite high concentration?\r\n- [ ] Checked if A publicly discloses suppliers (annual report) and B is absent\r\n\r\n### Contractor / competitor relationship\r\n- [ ] Checked if A and B are direct competitors in T (A buying from competitor = forced, unlikely)\r\n- [ ] Checked if A positions its product as a substitute for B's\r\n\r\n### Negative verification outcome\r\n- [ ] Negative evidence FOUND → reduce confidence 1-2 levels (or overturn if fundamental contradiction)\r\n- [ ] Negative search performed, NO evidence found → \"negative verification passed\" — supports inference\r\n- [ ] Negative search NOT possible → flag \"negative verification incomplete\"\r\n\r\n## Evidence Freshness Tracking (NEW v0.4.0)\r\n\r\n**Principle**: Supply chains change. A 2023 teardown does not prove a 2026 relationship. Every evidence item carries a timestamp.\r\n\r\n### Freshness levels\r\n| Level | Age | Treatment |\r\n|-------|-----|-----------|\r\n| **Current** | ≤12 months | Full weight |\r\n| **Moderate** | 12-18 months | Normal weight; flag in report |\r\n| **Stale** | >18 months | Down-weight one level; cannot alone support \"high\" |\r\n| **Unknown** | date not found | Treat as stale; flag prominently |\r\n\r\n### Freshness rules\r\n- [ ] Every evidence item in the report has a date or estimated date\r\n- [ ] If no date can be found, marked \"date unknown\" and treated as stale\r\n- [ ] Conclusion resting primarily on stale evidence → capped at \"medium\"\r\n- [ ] Corroborating leads that are all stale → overall capped at \"medium\"\r\n- [ ] Report includes a \"evidence freshness table\" listing each lead + evidence date + freshness level\r\n\r\n### Refreshing stale evidence\r\n- [ ] If stale evidence is critical, attempt to re-verify: search for newer teardowns, newer supplier disclosures, updated customer lists\r\n- [ ] Document whether the relationship appears current or historical\r\n\r\n## Competitive Triangulation Checklist\r\n\r\n### Peer identification\r\n- [ ] Identified 3-10 direct competitors / industry peers of A\r\n- [ ] Peer selection: same industry classification + similar product + similar revenue size\r\n\r\n### Gap verification\r\n- [ ] Peer C₁ patent portfolio checked for T → [has / partial / none]\r\n- [ ] Peer C₂ patent portfolio checked for T → [has / partial / none]\r\n- [ ] Peer Cₙ patent portfolio checked for T → [has / partial / none]\r\n\r\n### Convergence check\r\n- [ ] For peers with gaps, does T's ownership converge to the same B?\r\n- [ ] Any peers pointing to different owners → flag fragmented supply\r\n- [ ] **Quality-aware check (v0.4.0)**: does B's patent quality support the convergence? (Core → strong signal; Peripheral → weak signal)\r\n\r\n### Triangulation boost\r\n- [ ] ≥3 peers same gap → same B: **+2 confidence levels**\r\n- [ ] 2 peers same gap → same B: **+1 confidence level**\r\n- [ ] 1 peer same gap: note in report, no boost\r\n- [ ] Peers diverge: no boost, flag fragmentation\r\n- [ ] Triangulation cannot override trade secret cap or financial cross-validation cap\r\n\r\n## Financial Cross-Validation Checklist\r\n\r\n### Forward direction\r\n- [ ] Checked A's top-5 supplier list in annual report\r\n- [ ] Checked A's supplier concentration ratio (>30% = single-supplier risk)\r\n- [ ] Matched A's procurement categories against B's product lines\r\n- [ ] If B is listed: checked B's top-5 customer list for A-like profile\r\n- [ ] Checked B's customer concentration (>20% from single customer)\r\n- [ ] Checked accounts receivable / payable patterns for correlation\r\n- [ ] Checked related-party transaction disclosures\r\n- [ ] Checked A's procurement growth vs. B's revenue growth (correlation)\r\n\r\n### Reverse direction\r\n- [ ] Checked B's top-5 customer list\r\n- [ ] Checked B's revenue by industry segment\r\n- [ ] For each candidate A: checked A's supplier list for B\r\n- [ ] For each candidate A: checked A's related-party disclosures\r\n- [ ] Estimated revenue contribution of each candidate (where data allows)\r\n\r\n### Financial corroboration scoring\r\n- [ ] **Strong**: B named as A's supplier OR A named as B's customer in official filings\r\n- [ ] **Medium**: Procurement category + revenue scale matches\r\n- [ ] **None**: No financial data found → **cap overall confidence at \"medium\"**\r\n- [ ] **Negative**: Financial data contradicts → **reduce confidence by 1 level**\r\n\r\n## Counter-Intelligence Checklist\r\n\r\n### Patent-ownership side\r\n- [ ] Is B's patent ownership abnormally dispersed? (shell companies / affiliates holding separately)\r\n- [ ] Is there a shell company set up specifically to hold patents? (low registered capital, no real operations)\r\n- [ ] Are patent transfers frequent? (possibly to obscure true ownership)\r\n\r\n### A's patent-strategy side\r\n- [ ] Does A deliberately not file patents to hide its tech roadmap?\r\n- [ ] Do A's patent filings mismatch its actual product technology? (decoy filings)\r\n\r\n### Anomaly signals\r\n- [ ] Is A's supplier-disclosure abnormally terse or deliberately vague?\r\n- [ ] Are key technical staff employed via labor dispatch / outsourcing rather than directly?\r\n- [ ] Are there related entities with frequent changes of name / legal representative / shareholders?\r\n\r\nIf any anomaly exists → flag \"counter-intelligence risk\" in the report.\r\n\r\n## Trade Secret / Know-How Indicators\r\n\r\n### Indicator 1 · Industry norms\r\n- [ ] T is in a trade-secret-dominant industry: semiconductor process, chemical formulation, pharma manufacturing process, food/beverage formula, metallurgy, specialty coatings, advanced materials synthesis\r\n- [ ] Industry is known for \"black box\" secrecy (e.g., TSMC process, ASML optics, DuPont formulas)\r\n\r\n### Indicator 2 · Company NDA / confidentiality culture\r\n- [ ] A has public record of aggressive NDA enforcement or trade secret litigation\r\n- [ ] A's employee contracts include strict non-compete + confidentiality clauses\r\n- [ ] A's facilities known for tight access control / \"clean room\" culture\r\n- [ ] A's technical publications / conference presentations notably sparse or redacted\r\n\r\n### Indicator 3 · Patent filing pattern (strategic selectivity)\r\n- [ ] A files patents across many domains but conspicuously avoids filing in T\r\n- [ ] A's patent portfolio shows carefully scoped filings (protecting what can be reverse-engineered, hiding what cannot)\r\n- [ ] A's patent filings in T's general category exist but intentionally omit key implementation details\r\n\r\n### Indicator 4 · R&D intensity vs. patent output\r\n- [ ] A has significant R&D team / budget in T's domain (job postings, R&D headcount, R&D expense)\r\n- [ ] A produces few or zero patents in T despite the R&D investment\r\n- [ ] A's R&D-to-patent ratio in T is significantly lower than industry peers\r\n\r\n### Indicator 5 · Technology nature (protectability)\r\n- [ ] T is a process / method / formula / manufacturing technique (hard to reverse-engineer → good for trade secret)\r\n- [ ] T is a software algorithm embedded in a black-box system (source code hidden → trade secret feasible)\r\n- [ ] T would be difficult to prove infringement even if patented (→ low incentive to patent)\r\n\r\n### Indicator 6 · Trade secret litigation history\r\n- [ ] A has sued others for trade secret theft in T's domain\r\n- [ ] A's peers in the industry have active trade secret litigation\r\n- [ ] Industry news reports trade secret disputes related to T\r\n\r\n### Scoring\r\n- **4-6 indicators → High likelihood trade secret**: gap likely strategic; default \"no dependency\" unless procurement/teardown overrides\r\n- **2-3 indicators → Possible trade secret**: record as alternative hypothesis; proceed normally\r\n- **0-1 indicators → Trade secret unlikely**: gap likely reflects real dependency; proceed normally\r\n\r\n### Override signal\r\n- [ ] Lead 4 (procurement) returns high-confidence evidence of external sourcing → overrides trade secret hypothesis\r\n- [ ] Lead 5 (teardown) physically identifies third-party component → overrides trade secret hypothesis\r\n\r\n## Confidence Scoring Rules\r\n\r\n### Per-lead confidence (v0.4.0)\r\n\r\n| Lead | High | Medium | Low |\r\n|------|------|--------|-----|\r\n| Patent ownership + quality | B holds Core-quality T patents, valid timeline | B holds Supporting-quality patents, timeline OK | T patents found but peripheral quality / owner unclear |\r\n| Corporate affiliation | JV / mutual investment | Executive overlap | Same address / same legal rep (weak) |\r\n| Talent flow | Multiple B technical staff joined A | Single key person joined | Job postings only mention B background |\r\n| Procurement | B directly named as supplier | B's category mentioned | Category-related but no clear pointer |\r\n| Product teardown | Physically confirmed B component | Teardown report mentions but unconfirmed | No teardown report |\r\n\r\n**Evidence freshness adjustment (all leads)**: stale (>18 months) evidence down-weights one level.\r\n\r\n### Overall confidence (base)\r\n\r\n- **High**: ≥2 high-confidence leads corroborating\r\n- **Medium**: 1 high + 1 medium corroborating, or multiple medium leads partially corroborating\r\n- **Low**: only 1 lead hit, or leads contradict each other\r\n- **Cannot infer**: 0 leads hit\r\n\r\n### Adjustment layers (apply in order, each CAN cap or boost):\r\n\r\n**Layer 1 — Trade secret adjustment**:\r\n- Trade secret \"high\" + no procurement/teardown → cap at \"low\"\r\n- Trade secret \"high\" + procurement/teardown confirms external sourcing → no cap\r\n- Trade secret \"possible\" → no cap, record as alternative hypothesis\r\n\r\n**Layer 2 — Triangulation boost**:\r\n- ≥3 peers same gap → same B: **+2 levels**\r\n- 2 peers same gap → same B: **+1 level**\r\n- Cannot override Layer 1 cap\r\n\r\n**Layer 3 — Negative verification** (v0.4.0):\r\n- Negative evidence found → **reduce 1-2 levels** (overturn if fundamental contradiction)\r\n- Negative verification passed → no change\r\n- Negative verification incomplete → no change, flag\r\n\r\n**Layer 4 — Evidence freshness** (v0.4.0):\r\n- Conclusion rests primarily on stale evidence → **cap at \"medium\"**\r\n- All corroborating leads stale → **cap at \"medium\"**\r\n\r\n**Layer 5 — Financial cross-validation cap**:\r\n- Strong/medium corroboration → no cap\r\n- No corroboration → **cap at \"medium\"** (exception: procurement/teardown physical evidence)\r\n- Negative corroboration → **reduce by 1 level**\r\n\r\n### Final confidence = min(base ± adjustments, caps applied in order)\n\nFile v0.1.2:references/industry-guide.md\n\n# Industry-Specific Patent & Trade Secret Guide\r\n\r\nDifferent industries have fundamentally different patent/trade-secret dynamics. This guide provides vertical-specific signals, verification strategies, and common pitfalls.\r\n\r\n---\r\n\r\n## 1. Semiconductor\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Manufacturing processes** (lithography, etching, deposition, packaging): Overwhelmingly trade secrets. TSMC, Samsung, Intel do NOT patent their process recipes. The 6-indicator trade secret assessment will almost always score 4+.\r\n- **Chip architectures / designs** (ISA, microarchitecture, interconnects): Strong patent coverage. ARM, x86, RISC-V are heavily patented.\r\n- **EDA tools / design methodologies**: Mixed. Some patented, some trade-secret algorithms.\r\n\r\n### Key Signals\r\n- A foundry's process node advancement without corresponding patents → trade secret, not gap\r\n- Fabless companies (e.g., MediaTek, UNISOC) lacking process patents → expected, they don't manufacture\r\n- Equipment makers (e.g., ASML, AMAT, LAM) → strong patent portfolios, less trade secret reliance\r\n\r\n### Verification Strategy\r\n- **Teardown > everything else**: Chip teardowns (TechInsights, ICWorks) are the gold standard\r\n- **Patent gap in chip design**: meaningful signal for IP licensing (e.g., ARM architecture license)\r\n- **Patent gap in manufacturing**: almost always trade secret, do not infer supplier dependency\r\n- **Financial cross-validation**: check foundry customer concentration in annual reports\r\n\r\n### Common A-Share Examples\r\n- SMIC (中芯国际): Process patents are sparse vs. TSMC; this reflects technology gap + trade secret strategy, not external sourcing\r\n- Hygon (海光信息): x86 license from AMD → genuine IP dependency\r\n- Cambricon (寒武纪): AI chip architecture patents → strong internal IP\r\n\r\n---\r\n\r\n## 2. Pharmaceuticals\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Compound patents** (small molecule structures): Heavily patented. The molecule IS the product.\r\n- **Manufacturing processes** (synthesis routes, purification, formulation): Often trade secrets. Same drug, different manufacturers may use different routes.\r\n- **Clinical trial data**: Regulatory exclusivity + trade secrets, not patents.\r\n- **Biologics / Biosimilars**: Complex mix. Cell line development and culture conditions are trade secrets.\r\n\r\n### Key Signals\r\n- Drug A has expired compound patent but no generic versions → manufacturing process is the barrier (trade secret or new process patent)\r\n- Biologic has zero process patents → expected, cell culture conditions are trade secrets\r\n- ANDA (generic drug application) filings → reveals which companies are trying to enter\r\n\r\n### Verification Strategy\r\n- **Patent expiry + generic entry timing**: The most reliable signal\r\n- **DMF (Drug Master File) filings**: Companies filing DMFs for an API are the actual suppliers\r\n- **CDE (药审中心) registration data**: China-specific regulatory filings\r\n- **Procurement/tender data**: Hospital procurement records, centralized drug procurement (集采)\r\n\r\n### Common A-Share Examples\r\n- Hengrui (恒瑞医药): Strong compound + formulation patents → internal R&D driven\r\n- WuXi AppTec (药明康德): CDMO, process patents + trade secrets → client dependency analysis\r\n- Generic manufacturers (Huahai 华海, Hisun 海正): Watch ANDA filing pipeline for target identification\r\n\r\n---\r\n\r\n## 3. New Energy / Battery\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Battery chemistry** (cathode materials, electrolyte formulations): Strong patent coverage. CATL, LG Energy, Panasonic all file aggressively.\r\n- **Manufacturing process** (coating, stacking, formation): Mixed. Some patented equipment designs, some process know-how.\r\n- **BMS (Battery Management System)**: Software-heavy, patent + trade secret mix.\r\n- **Recycling processes**: Emerging patent landscape, rapidly evolving.\r\n\r\n### Key Signals\r\n- Battery maker without cathode material patents → likely sourcing from material suppliers\r\n- EV maker without battery patents → definitely sourcing from battery makers (BYD is the exception — vertically integrated)\r\n- Separator / electrolyte additive patents → strong signal of capability depth\r\n\r\n### Verification Strategy\r\n- **Supply agreement announcements**: Battery supply contracts are often publicly announced (e.g., CATL → Tesla/NIO/XPeng)\r\n- **Product certification**: EV models must list battery suppliers for regulatory approval\r\n- **Capacity expansion + customer matching**: When B expands capacity and A launches new models → timing correlation\r\n- **Teardown**: EV battery pack teardowns explicitly identify cell manufacturers\r\n\r\n### Common A-Share Examples\r\n- CATL (宁德时代): Dominant patent portfolio, easily identified as supplier via company announcements\r\n- EV makers (NIO 蔚来, XPeng 小鹏): Zero cell patents → 100% external sourcing\r\n- Putailai (璞泰来), Tianci (天赐材料): Material suppliers → use reverse workflow (their customers are battery makers)\r\n\r\n---\r\n\r\n## 4. Consumer Electronics\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Industrial design / appearance**: Strong design patent coverage\r\n- **Internal components**: Component selection is procurement-driven, not IP-driven. Apple doesn't patent the OLED panel itself.\r\n- **Assembly / integration**: Trade secret (Foxconn's assembly process)\r\n- **Software / UX**: Copyright + trade secret + some patents\r\n\r\n### Key Signals\r\n- **Patent gap is THE NORM in consumer electronics**: A company not patenting components it buys is expected behavior. A patent gap here is weak signal.\r\n- Component teardowns (iFixit, etc.) → the most reliable data source\r\n- Supplier list in annual report → Apple, Xiaomi, Huawei all publish supplier lists\r\n\r\n### Verification Strategy\r\n- **Teardown > supplier list > patent gap**: Reverse priority vs. semiconductor/pharma\r\n- Patent gap in consumer electronics is rarely a meaningful signal — companies consciously don't patent purchased components\r\n- Focus on supplier lists, certification filings (3C certification in China), and teardowns\r\n- **Trade secret assessment**: almost always \"trade secret unlikely\" → but this doesn't mean supply-chain dependency is confirmed; it means patent gap is the wrong signal here\r\n\r\n### Common A-Share Examples\r\n- Xiaomi (小米): Buys panels, chips, cameras → patent gap expected\r\n- Luxshare (立讯精密), Goertek (歌尔股份): Contract manufacturers → their customers are the brand companies (reverse workflow)\r\n- BOE (京东方): Panel maker → use reverse workflow to identify downstream phone/TV brand customers who buy from BOE\r\n\r\n---\r\n\r\n## 5. Materials / Chemicals\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Formulations / recipes**: Classic trade secret domain. Think Coca-Cola, DuPont's Kevlar, 3M's adhesives.\r\n- **Synthesis routes**: Mixed. Some patented (process patents), some trade secrets.\r\n- **Application patents**: How to USE a material → often patented by the user, not the producer. This creates a useful \"reverse\" patent gap signal.\r\n\r\n### Key Signals\r\n- Material user A has application patents for material M but zero production patents for M → A probably buys M\r\n- Material producer B has composition + process patents → strong signal of capability\r\n- High R&D-to-patent ratio → typical for materials companies, not suspicious\r\n\r\n### Verification Strategy\r\n- **Customer audits / qualifications**: Materials for electronics/auto/aerospace require lengthy qualification (1-3 years). Once qualified, switching costs are enormous.\r\n- **Capacity expansion alignment**: When B adds capacity and A enters new markets → timing correlation\r\n- **Trade secret assessment**: Often scores high, requiring procurement/teardown evidence to draw conclusions\r\n- **Financial cross-validation**: Check A's raw material procurement categories against B's product line\r\n\r\n### Common A-Share Examples\r\n- Shin-Etsu (信越化学), SUMCO: Silicon wafer suppliers → their customers are semiconductor fabs (reverse workflow)\r\n- Weihua (万华化学): MDI patents → downstream polyurethane users as customers\r\n- Rongbai (容百科技), Dangsheng (当升科技): Cathode material suppliers → customers are battery makers\r\n\r\n---\r\n\r\n## Industry Selection Quick Reference\r\n\r\n| If the target is... | Priority verification lead | Patent gap signal strength | Trade secret risk |\r\n|---------------------|---------------------------|---------------------------|-------------------|\r\n| Fabless semiconductor | Patent ownership (licensing) | **Medium** | Low |\r\n| Foundry / process | Teardown + financial | **Weak** — process is trade secret | High |\r\n| Innovative pharma | Patent expiry + ANDA | **Strong** — compounds are patented | Low |\r\n| CDMO / CMO | Procurement + financial | **Weak** — processes are trade secrets | High |\r\n| EV maker | Teardown + announcements | **Strong** — battery cells are outsourced | Low |\r\n| Battery maker | Reverse workflow | **Medium** | Medium |\r\n| Consumer electronics OEM | Teardown + supplier list | **Weak** — patent gap is normal | Low |\r\n| Materials producer | Reverse workflow | **Medium** | High |\n\nFile v0.1.2:references/workflow.md\n\n# Detailed Workflow\r\n\r\n## Step 1: Trigger lead identification + direction selection\r\n\r\n**Goal**: Define the analysis target, determine analysis direction, and identify a potential patent gap.\r\n\r\n**Operations**:\r\n1. Confirm target company (full name, ticker)\r\n2. Determine analysis direction based on user intent:\r\n   - **Forward**: user asks \"who supplies X?\" / \"identify suppliers\" / \"what company provides the tech?\" → target is A (the patent-gap company)\r\n   - **Reverse**: user asks \"who are B's customers?\" / \"which downstream companies depend on this?\" / \"what companies need this technology?\" → target is B (the patent owner)\r\n3. Identify the key technology T and/or product P:\r\n   - If not specified by user, extract from product docs, annual report, prospectus, research reports\r\n4. Preliminarily judge patent status:\r\n   - Forward: does A likely lack T's patents?\r\n   - Reverse: does B likely hold dominant/leading T patents?\r\n\r\n**Output**: Direction (forward/reverse) + target company + T + P + preliminary judgment\r\n\r\n## Step 2A: Patent gap confirmation (forward direction)\r\n\r\n**Goal**: Confirm with data whether A truly lacks T-related patents, and rule out legitimate-use scenarios.\r\n\r\n**Operations**:\r\n1. Search A's patents using patent search tools (tushare patent API / PatSnap)\r\n   - Dimensions: A's company name + technology T keywords + IPC class\r\n   - Scope: CN patents + US patents + PCT (per A's market coverage)\r\n   - Note: A may hold patents via subsidiaries / affiliates — search A's related entities\r\n2. Determine gap:\r\n   - A (incl. affiliates) has zero T-related patents → gap confirmed\r\n   - A has T-related patents but count significantly below industry average → partial gap\r\n   - A has sufficient T-related patents → no gap, terminate\r\n3. Rule out legitimate-use scenarios (if gap confirmed):\r\n   - Cross-licensing, acquisition inheritance, patent pool, open-source\r\n\r\n**Output**: Gap determination + excluded legitimate-use scenarios\r\n\r\n## Step 2B: Patent dominance confirmation (reverse direction)\r\n\r\n**Goal**: Confirm that B indeed holds a dominant position in T's patent landscape.\r\n\r\n**Operations**:\r\n1. Search T's patent landscape comprehensively\r\n2. Rank patent holders by quantity and citation count\r\n3. Determine B's dominance:\r\n   - B holds >30% of T's patents OR top-3 cited in T → dominant position confirmed\r\n   - B holds 10-30% → moderate position\r\n   - B holds <10% → weak position, may not be the key technology gatekeeper\r\n4. Identify B's patent family: core patents + timing + jurisdiction coverage\r\n\r\n**Output**: Dominance level (dominant / moderate / weak) + B's core patent list\r\n\r\n## Step 3: Trade secret / know-how assessment\r\n\r\n**Goal**: Determine whether the patent gap reflects true external dependency or deliberate trade secret / know-how protection.\r\n\r\n**Core distinction**:\r\n- **Trade secret**: A has the technology internally, but protects it through secrecy\r\n- **True gap → dependency**: A genuinely lacks the technology and must source it externally\r\n\r\n**Trade secret assessment — 6 indicators**:\r\n\r\n| # | Indicator | Strong trade secret signal | Strong dependency signal |\r\n|---|-----------|---------------------------|--------------------------|\r\n| 1 | Industry norms | Semiconductor process, chemical formula, pharma manufacturing, metallurgy | Consumer electronics, commoditized components |\r\n| 2 | NDA / confidentiality culture | Aggressive NDAs, non-compete lawsuits, \"black box\" secrecy | Open publications, transparent R&D |\r\n| 3 | Patent filing pattern | Files patents in some domains, conspicuously avoids T | Files patents evenly or rarely |\r\n| 4 | R&D intensity vs. patent output | High R&D spend + low T patents | Little to no R&D in T's domain |\r\n| 5 | Technology nature | Process/method/formula (hard to reverse-engineer) | Physical component (identifiable in teardown) |\r\n| 6 | Trade secret litigation | Company/industry trade secret theft lawsuits | Patent litigation is the norm |\r\n\r\n**Scoring**: 4-6 = high likelihood, 2-3 = possible, 0-1 = unlikely.\r\n\r\n**Override rule**: Procurement or teardown evidence can override trade secret finding.\r\n\r\n**Note for reverse workflow**: In reverse direction, spend LESS effort on trade secret assessment for each downstream A — the key question is whether B's patents are essential, not whether A could develop internally. Assessment still runs but reduced weight.\r\n\r\n**Output**: Trade secret likelihood + indicator breakdown\r\n\r\n## Step 4: Multi-lead parallel verification\r\n\r\n**Goal**: Activate 5 leads in parallel, each with an independent confidence score. **Every lead must record its evidence timestamp** (evidence freshness, v0.4.0). Detailed scoring rules in `checklists.md`.\r\n\r\n### Lead 1 · Patent ownership + quality (confidence: medium, adjusted by quality)\r\n- Forward: Search T's patent holders, identify primary owner B\r\n- Reverse: For each candidate A, search whether A's products use T and whether A holds T patents\r\n- Check patent status, timeline, NPE risk\r\n- **Patent quality scoring (v0.4.0)**: Do not count patents — score them. See `checklists.md` for the 4-factor quality score:\r\n  - Claim breadth (independent claims, scope)\r\n  - Citation impact (forward citations, whether later patents build on it)\r\n  - Family size (PCT/multi-jurisdiction vs single-country)\r\n  - Legal status (granted & maintained vs lapsed/revoked/pending)\r\n  - Quality levels: Core (≥8/12), Supporting (4-7/12), Peripheral (0-3/12)\r\n- Quality-adjust Lead 1 confidence: Core patents → confidence can reach high; Peripheral patents only → cap at low\r\n\r\n### Lead 2 · Corporate affiliation (confidence: high)\r\n- Search A-B affiliation via Tianyancha / Qichacha / Qixinbao\r\n- Check: mutual investment, JV, executive overlap, historical changes\r\n\r\n### Lead 3 · Talent flow (confidence: medium)\r\n- Search B→A technical-hire / job-change records\r\n- Use WebSearch + job sites\r\n- Multiple technical staff → high likelihood\r\n\r\n### Lead 4 · Procurement (confidence: high)\r\n- Forward: Search A's tender announcements, supplier disclosures for B\r\n- Reverse: Search B's listed customers, annual report client disclosures\r\n- Directly named → high; category only → medium\r\n\r\n### Lead 5 · Product teardown (confidence: very high)\r\n- Physical evidence of B components in A's products\r\n- Teardown reports / BOM analysis / product certifications\r\n\r\n**Evidence freshness (v0.4.0, applies to ALL leads)**:\r\n- Record the date of each evidence item\r\n- Freshness levels:\r\n  - **Current** (≤12 months): full weight\r\n  - **Moderate** (12-18 months): normal weight, flag in report\r\n  - **Stale** (>18 months): down-weight one level; cannot alone support \"high\" confidence\r\n  - **Unknown date**: treat as stale, flag prominently\r\n- A conclusion resting primarily on stale evidence is capped at \"medium\"\r\n\r\n**Output**: Per-lead hit status + per-lead confidence + patent quality score (Lead 1) + evidence timestamps\r\n\r\n## Step 5: Competitive triangulation\r\n\r\n**Goal**: Check whether A's industry peers show the same patent gap pattern, providing cross-company corroboration.\r\n\r\n**Why this matters**: A single company's patent gap could be random or internal. But if 3-5 industry peers ALL lack T's patents and ALL point to the same technology owner B, the pattern is unlikely to be coincidental.\r\n\r\n**Operations**:\r\n1. Identify A's direct competitors / industry peers (C₁, C₂...Cₙ): 3-10 peers, same industry classification, similar product/revenue profile\r\n2. For each peer Cᵢ, check patent gap for T\r\n3. For peers with gaps, check where T likely comes from (same B or different owners)\r\n4. **Quality-aware triangulation (v0.4.0)**: If multiple peers converge on B AND B's patents score as Core quality → strongest possible signal. If B's patents are only Peripheral, convergence is weaker (B may be a paper tiger holding low-value patents).\r\n\r\n| Pattern | Interpretation | Triangulation boost |\r\n|---------|---------------|---------------------|\r\n| ≥3 peers same gap → same B, B=Core quality | Industry-wide dependency on genuine tech leader | **+2 levels** |\r\n| ≥3 peers same gap → same B (any quality) | Industry-wide dependency | **+2 levels** |\r\n| ≥2 peers same gap → same B | Significant corroboration | **+1 level** |\r\n| 1 peer same gap → B | Mild corroboration | No boost |\r\n| Peers point to different owners | Fragmented supply → B weaker | No boost, flag |\r\n| Peers have T patents | A is the outlier → investigate | No boost |\r\n\r\n**Output**: Peer gap matrix + triangulation boost + quality-aware interpretation\r\n\r\n## Step 6: Negative verification (NEW v0.4.0)\r\n\r\n**Goal**: Actively attempt to DISPROVE the supply-chain inference. An inference that survives active disproof is far more credible than one that was never tested.\r\n\r\n**Core principle**: Positive evidence (patents, affiliations, teardowns) can be coincidental or over-interpreted. Negative verification searches for the \"why not\" — evidence that B is NOT A's supplier despite the positive signals.\r\n\r\n**Operations** — search each of the following for counter-evidence:\r\n\r\n1. **Revenue/scale mismatch**:\r\n   - Does B's reported revenue exceed what A could plausibly buy? (if B's revenue is 100x A's procurement budget for T, B may not be A's supplier — or A is trivial for B)\r\n   - Does B's capacity announcement target products A doesn't make?\r\n\r\n2. **Product-category mismatch**:\r\n   - Does B's product line actually include what A needs? (B holds T patents but may not manufacture the specific product variant A uses)\r\n   - Check B's product catalog / website for the specific component A needs\r\n\r\n3. **Geographic/logistics infeasibility**:\r\n   - Are A and B in incompatible geographies for the component type? (some components are regional by nature — e.g., regulatory, shipping cost)\r\n   - Check trade routes, tariffs, local-content requirements\r\n\r\n4. **Explicit customer-list exclusion**:\r\n   - If B publicly lists customers (annual report top-5, website case studies, press releases), does it explicitly EXCLUDE A?\r\n   - If B's top-5 customer list is disclosed and A is not among them AND B's concentration is high → strong negative evidence\r\n\r\n5. **Contractor/competitor relationship**:\r\n   - Is A actually a competitor of B in T? (A wouldn't buy from its direct competitor unless forced)\r\n   - Is A's product positioned as a substitute for B's?\r\n\r\n6. **Patent-licensing-only alternative**:\r\n   - Does A license B's patents but buy the physical product from a THIRD party (C)? If so, B is a licensor, not a supplier — the supply chain passes through C.\r\n\r\n**Negative verification outcome**:\r\n\r\n| Outcome | Impact on confidence |\r\n|---------|---------------------|\r\n| Negative evidence FOUND (any of the above confirmed) | **Reduce confidence by 1-2 levels**, flag the specific contradiction. If the contradiction is fundamental (e.g., A is B's direct competitor), the inference may be overturned. |\r\n| Negative search performed, no evidence found | **Confidence confirmed** — record \"negative verification passed\" in report; supports high confidence |\r\n| Negative search could not be performed (data unavailable) | Note in report; confidence unchanged but flagged as \"negative verification incomplete\" |\r\n\r\n**Critical rule**: The report MUST document that negative verification was attempted. A report without negative verification is incomplete.\r\n\r\n**Output**: Negative evidence found/not found + specifics + impact on confidence\r\n\r\n## Step 7: Multi-lead fusion + confidence scoring\r\n\r\n**Goal**: Cross-validate and produce an overall determination.\r\n\r\n**Fusion rules**:\r\n- ≥2 high-confidence leads corroborating → overall \"high\", can conclude\r\n- 1 high + 1 medium corroborating → overall \"medium\"\r\n- Only 1 lead hit, no corroboration → overall \"low\", hypothesis only\r\n- 0 leads hit → terminate, cannot infer\r\n\r\n**Trade secret override rules**:\r\n- Trade secret \"high\" + no procurement/teardown → cap at \"low\"\r\n- Trade secret \"high\" + procurement/teardown confirms → no cap\r\n\r\n**Triangulation boost** (applied after base scoring):\r\n- +2 levels if ≥3 peers confirm same gap → same B\r\n- +1 level if ≥2 peers confirm same gap → same B\r\n- Triangulation cannot override trade secret cap\r\n\r\n**Negative verification adjustment (v0.4.0)**:\r\n- Negative evidence found → reduce by 1-2 levels (or overturn if fundamental contradiction)\r\n- Negative verification passed → no change (recorded as supporting)\r\n- Negative verification incomplete → no change, flag in report\r\n\r\n**Evidence freshness adjustment (v0.4.0)**:\r\n- If the conclusion rests primarily on stale (>18 months) evidence → cap at \"medium\"\r\n- Each lead using stale evidence is down-weighted one level before fusion\r\n- If the only corroborating leads are stale → overall capped at \"medium\"\r\n\r\n**Role determination**:\r\n- Leads 4/5 hit (procurement / teardown) → B is A's direct supplier (forward) / A is B's customer (reverse)\r\n- Only leads 1/2 hit → technology licensor or indirect affiliation\r\n- Lead 1 hit but B is NPE → B licenses only, not in supply chain\r\n\r\n**Output**: Overall confidence + all adjustments applied + B's/A's role\r\n\r\n## Step 8: Financial cross-validation\r\n\r\n**Goal**: Cross-validate the inferred supply-chain relationship against hard financial data.\r\n\r\n**Operations** (select applicable checks based on direction):\r\n\r\n### Forward direction (A → find supplier B):\r\n1. A's supplier concentration (top-5 supplier list, concentration ratio)\r\n2. A's procurement category matching (does A's procurement align with B's products?)\r\n3. B's customer concentration (if listed: top-5 customers, concentration)\r\n4. Accounts receivable / payable patterns, related-party disclosures\r\n\r\n### Reverse direction (B → find customers A):\r\n1. B's customer concentration, revenue by industry segment\r\n2. For each candidate A: top-5 supplier list, related-party disclosures\r\n\r\n### Financial corroboration scoring:\r\n\r\n| Finding | Corroboration | Confidence effect |\r\n|---------|---------------|-------------------|\r\n| B named as A's supplier / A as B's customer | **Strong** | No cap, confirms |\r\n| Procurement category + revenue size matches | **Medium** | Supports |\r\n| Related-party disclosures confirm | **Strong** | Confirms |\r\n| No financial evidence | **None** | Cap at \"medium\" |\r\n| Financial data contradicts | **Negative** | Reduce 1 level, flag |\r\n\r\n**Output**: Financial corroboration level + specific evidence + confidence cap applied\r\n\r\n## Step 9: Alternative-hypothesis check\r\n\r\nSee `checklists.md`. Explicitly exclude alternative explanations for each inference.\r\n\r\n## Step 10: Counter-intelligence check\r\n\r\nSee `checklists.md`. Check for shell companies / dispersed patents / decoy filings.\r\n\r\n## Step 11: Output report\r\n\r\nOrganize per the I/O contract in SKILL.md. Must include:\r\n- Analysis direction (forward / reverse)\r\n- Relationship determination (incl. role)\r\n- Trade secret assessment result\r\n- Evidence chain (5-lead hit table + confidence + **patent quality scores** + **evidence timestamps**)\r\n- Triangulation result (peer gap matrix + boost applied)\r\n- **Negative verification result** (searched + found/not found + impact)\r\n- Financial corroboration (evidence + level)\r\n- Overall confidence + rationale (all adjustments transparent: trade secret, triangulation, negative verification, evidence freshness, financial)\r\n- Alternative-hypothesis list\r\n- Counter-intelligence risk flag\r\n- Follow-up verification suggestions\r\n\r\n---\r\n\r\n# Appendix: Reverse Workflow (Supplier → Clients)\r\n\r\n## When to use\r\n\r\nTrigger when the user provides a technology-owning company B and wants to identify downstream customers. Common scenarios:\r\n\r\n- \"台积电的先进封装客户有哪些？\"\r\n- \"Qualcomm's 5G modem customers besides Apple?\"\r\n- \"哪些车企用了宁德时代的麒麟电池？\"\r\n- \"What companies depend on ASML's EUV lithography?\"\r\n\r\n## Process\r\n\r\n### R1: Confirm B's patent dominance\r\n- Run Step 2B to validate B's position in T\r\n- **Apply patent quality scoring** to B's core patents — dominance + Core quality = genuine gatekeeper; dominance + Peripheral quality = weak gatekeeper\r\n- If B is not dominant → warn and proceed at reduced confidence\r\n\r\n### R2: Identify technology T and its necessity\r\n- What exactly does B's patented technology enable?\r\n- Is T a must-have (essential standard) or a nice-to-have (one of many options)?\r\n- Essential standard → downstream dependency is automatic; non-essential → need product-level evidence\r\n\r\n### R3: Candidate screening\r\n- Identify all companies whose products plausibly need T\r\n- Build candidate list: A₁, A₂...Aₙ\r\n\r\n### R4: For each high-priority candidate, run abbreviated forward workflow\r\n- Step 2A: Confirm Aᵢ lacks T patents\r\n- Step 3: Quick trade secret check (reduced weight)\r\n- Step 4: Run 5 leads for Aᵢ→B (with evidence timestamps)\r\n- Step 6: Negative verification per candidate\r\n- Step 8: Financial cross-validation (check B's customer list for Aᵢ)\r\n\r\n### R5: Aggregate report\r\n- List all confirmed customers with per-customer confidence\r\n- Map B's customer concentration risk\r\n- Flag any unnatural customer concentration (>30% from single customer)\n\nFile v0.1.2:CHANGELOG.md\n\n# Changelog\r\n\r\nAll notable changes to this skill are documented here.\r\nFormat based on [Keep a Changelog](https://keepachangelog.com/), adheres to [Semantic Versioning](https://semver.org/).\r\n\r\n## [0.4.0] - 2026-08-06\r\n\r\n### Added\r\n- **Patent quality scoring** (Step 4 / Lead 1): 4-factor quality score — claim breadth, citation impact, family size, legal status (0-12 scale). Core (≥8/12) / Supporting (4-7) / Peripheral (0-3) levels. Core patents allow Lead 1 to reach high confidence; peripheral-only caps at low. Quality-aware triangulation: convergence on a Core-quality holder is a stronger signal than convergence on a peripheral holder.\r\n- **Negative verification** (Step 6): Active disproof workflow. Search 5 counter-evidence classes — revenue/scale mismatch, product-category mismatch, geographic/logistics infeasibility, explicit customer-list exclusion, contractor/competitor relationship. Negative evidence found → reduce confidence 1-2 levels (or overturn); negative verification passed → supports inference; incomplete → flagged.\r\n- **Evidence freshness tracking**: Every lead carries an evidence timestamp. Freshness levels: Current (≤12mo, full weight) / Moderate (12-18mo, flag) / Stale (>18mo, down-weight one level) / Unknown (treat as stale). Conclusions resting primarily on stale evidence capped at \"medium\". Report includes evidence freshness table.\r\n\r\n### Changed\r\n- Workflow: 10 → 11 steps (new Step 6: Negative verification; subsequent steps renumbered)\r\n- Lead 1 renamed to \"Patent ownership + quality\" with quality-adjusted confidence\r\n- Fusion rules now include 5 adjustment layers: trade secret → triangulation → negative verification (new) → evidence freshness (new) → financial cap\r\n- Per-lead confidence table updated with patent quality and freshness dimensions\r\n- Triangulation now quality-aware (convergence on Core-quality holder = strongest signal)\r\n- Reverse workflow appendix: R1 applies patent quality scoring to B's dominance check; R4 adds negative verification per candidate\r\n- Key Principles: added patent-quality, negative-verification, and evidence-freshness principles\r\n- Frontmatter description and trigger keywords expanded (专利质量, 负向验证, patent quality, negative verification)\r\n- Tool Integration table: added negative verification, evidence freshness rows\r\n- I/O contract: added patent quality score, evidence timestamp, negative verification result fields\r\n- README.md rewritten for v0.4.0\r\n- `examples/sample-output.md`: updated to full 11-step format with quality scoring, negative verification, freshness table\r\n\r\n## [0.3.0] - 2026-07-21\r\n\r\n### Added\r\n- **Competitive triangulation** (Step 5): Cross-verify A's patent gap against 3-10 industry peers. 3+ peers same gap → same B = +2 confidence boost. Triangulation checklist in `references/checklists.md`.\r\n- **Financial cross-validation** (Step 7): Mandatory financial corroboration. Checks: top-supplier disclosure, customer concentration, AR/AP patterns, related-party transactions. No financial corroboration → confidence capped at \"medium\". Checklist in `references/checklists.md`.\r\n- **Reverse workflow** (supplier → clients): Start from technology owner B, identify downstream customers. Full appendix in `references/workflow.md` (R1-R5 process).\r\n- **Industry-specific guide** (`references/industry-guide.md`): 5 verticals — semiconductor, pharma, new energy/battery, consumer electronics, materials/chemicals. Each with key signals, verification strategy, and A-share examples.\r\n- **Real case study** (`examples/real-case.md`): Goertek (歌尔股份) & Apple VR supply chain. Demonstrates triangulation across 4 peers, financial cross-validation of 46.7% customer concentration, and ODM business model detection.\r\n\r\n### Changed\r\n- Workflow: 8 → 10 steps. Step 5: triangulation, Step 7: financial cross-validation. Subsequent steps renumbered.\r\n- Direction selection added to Step 1 (forward/reverse).\r\n- Step 2 split into 2A (gap) and 2B (patent dominance for reverse).\r\n- Confidence scoring: 3-layer adjustment (trade secret → triangulation → financial cap).\r\n- I/O contract, Tool Integration, Key Principles, frontmatter, trigger keywords all expanded.\r\n- README.md rewritten for v0.3.0.\r\n- `examples/sample-output.md` updated with triangulation and financial sections.\r\n\r\n## [0.2.0] - 2026-07-20\r\n\r\n### Added\r\n- **Trade secret / know-how assessment** (Step 3): New mandatory step after patent gap confirmation that evaluates whether the gap reflects deliberate trade secret protection rather than external dependency\r\n- 6-indicator trade secret assessment framework in `references/workflow.md`: industry norms, company NDA culture, patent filing pattern, R&D-to-patent ratio, technology nature, trade secret litigation history\r\n- Trade secret indicators checklist in `references/checklists.md` with scoring (4-6 = high likelihood, 2-3 = possible, 0-1 = unlikely)\r\n- Trade secret override rules: procurement (Lead 4) or teardown (Lead 5) findings with high-confidence physical evidence can override a trade secret finding\r\n- Trade secret adjustments to overall confidence scoring in `references/checklists.md` (confidence capped at \"low\" when trade secret likely and no physical sourcing evidence)\r\n\r\n### Changed\r\n- Workflow expanded from 7 to 8 steps to accommodate trade secret assessment\r\n- All subsequent step numbers updated (old Step 3→4, Step 4→5, Step 5→6, Step 6→7, Step 7→8)\r\n- Multi-lead fusion rules now incorporate trade secret override logic\r\n- Alternative-hypothesis checklist now explicitly includes trade secret / know-how protection\r\n- Key Principles updated with trade-secret-first rule and procurement/teardown override logic\r\n- Frontmatter description updated to mention trade secret distinction\r\n- README.md Chinese and English sections updated to document new trade secret capability\r\n- Sample output (`examples/sample-output.md`) updated with Step 3 trade secret assessment section\r\n\r\n### Design rationale\r\n- Patents require public disclosure; trade secrets do not. Many deep-tech companies (semiconductor processes, chemical formulations, pharma manufacturing) deliberately avoid patents to protect know-how. A patent gap therefore has two possible explanations: (a) true external dependency, or (b) deliberate trade secret strategy. Without explicitly evaluating this distinction, the skill risked false-positive supply-chain inferences.\r\n\r\n## [0.1.0] - 2026-07-09\r\n\r\n### Added\r\n- Initial release of patent-gap-supply-chain skill\r\n- 7-step workflow: trigger identification → patent gap confirmation → 5-lead parallel verification → fusion + confidence scoring → alternative-hypothesis check → counter-intelligence check → report output\r\n- Bilingual trigger keywords (Chinese + English) in description for cross-locale agent triggering\r\n- Applicability pre-check (Step 0): hardware / manufacturing / pharma / materials = strong; software / internet = weak; business-model innovation = not applicable\r\n- `references/workflow.md`: detailed per-step operations with decision rules and tool invocation\r\n- `references/checklists.md`: alternative-hypothesis checklist, counter-intelligence checklist, confidence scoring rules\r\n- `examples/sample-output.md`: worked example showing the full report format for a fictional case\r\n- Tool integration table covering tushare, PatSnap, Tianyancha, Qichacha, westock, WebSearch\n\nFile v0.1.2:examples/real-case.md\n\n# Real Case Study: Goertek (歌尔股份) & Apple's VR Supply Chain\r\n\r\nA worked example using a real Chinese A-share company. This demonstrates the full v0.3.0 workflow including competitive triangulation and financial cross-validation.\r\n\r\n**Disclaimer**: This case study is for educational demonstration of the skill workflow. All data is from public sources. This is NOT investment advice.\r\n\r\n---\r\n\r\n## Input\r\n\r\n- **Direction**: Forward (patent gap → find supplier relationships)\r\n- **Target A**: Goertek Inc. (歌尔股份, 002241.SZ)\r\n- **Technology T**: VR/AR optical and acoustic modules\r\n- **Background**: Goertek is known as an Apple supplier for AirPods and reportedly VR headsets. But does Goertek develop the core optical/acoustic tech in-house, or source it?\r\n\r\n---\r\n\r\n## Step 0: Applicability\r\n\r\nConsumer electronics / precision manufacturing → **Strong applicability**. Reference `industry-guide.md` consumer electronics section: patent gap in this sector is common; teardown and supplier lists take priority over patent gaps.\r\n\r\n---\r\n\r\n## Step 1: Trigger + Direction\r\n\r\n- Direction: Forward\r\n- A = Goertek (歌尔股份, 002241.SZ)\r\n- P = VR/AR headset components (optical modules, acoustic modules, haptics)\r\n- T = VR optical design, spatial audio algorithms, MEMS microphone arrays\r\n\r\n---\r\n\r\n## Step 2: Patent gap confirmation\r\n\r\nSearched Goertek's patent portfolio (CN + US):\r\n- Goertek holds **6,800+ patents** total, but these concentrate in:\r\n  - MEMS microphone manufacturing (their historical core business)\r\n  - Precision molding / injection processes\r\n  - Acoustic testing equipment\r\n- VR/AR optical design: **~80 patents** — mostly assembly methods and testing, NOT core optical design (lens recipes, waveguide design, display calibration)\r\n- Spatial audio algorithms: **~15 patents** — mostly implementation, not core algorithm IP\r\n\r\n**Gap determination**: **Partial gap** — Goertek patents cover manufacturing/assembly, but core optical design and audio algorithms appear sparse.\r\n\r\n---\r\n\r\n## Step 3: Trade secret assessment\r\n\r\n| Indicator | Assessment | Signal |\r\n|-----------|------------|--------|\r\n| 1. Industry norms | Consumer electronics manufacturing — not trade-secret-heavy | Neutral |\r\n| 2. NDA culture | Apple imposes strict NDAs on all suppliers; Goertek operates under Apple's secrecy regime | Trade secret* |\r\n| 3. Patent filing pattern | Goertek files heavily in MEMS/acoustics but lightly in VR optics → strategic selectivity possible | Trade secret |\r\n| 4. R&D vs. patent output | High R&D (6% of revenue), but VR optical patent output is low | Trade secret |\r\n| 5. Technology nature | Optical modules contain both physical components (identifiable) and calibration firmware (hidden) | Neutral |\r\n| 6. Trade secret litigation | No known trade secret litigation involving Goertek | Neutral |\r\n\r\n*Note: Apple's NDA is a special case — the secrecy is imposed by the customer, not Goertek's own choice. This is not a typical \"trade secret strategy.\"\r\n\r\n**Trade secret likelihood**: 2/6 indicators → **Possible trade secret**. Record as alternative hypothesis.\r\n\r\n---\r\n\r\n## Step 4: Multi-lead parallel verification\r\n\r\n| Lead | Hit | Detail | Confidence |\r\n|------|-----|--------|------------|\r\n| 1. Patent ownership | Partial | Key VR optical patents held by Apple (USPTO), Meta (Oculus), Sony; MEMS patents held by Knowles, Infineon | Medium |\r\n| 2. Corporate affiliation | No | No JV or investment relationship with Apple beyond standard supplier contract | — |\r\n| 3. Talent flow | Yes | Goertek hires optical engineers from Apple, Meta; but flow is FROM Apple TO Goertek, suggesting Apple is the tech source | Medium |\r\n| 4. Procurement | No | Goertek's supplier list is not publicly detailed; Apple imposes confidentiality | — |\r\n| 5. Product teardown | Yes | iFixit teardown of Apple Vision Pro shows multiple Goertek-manufactured components; Apple-designed chips (R1, M2) and sensors | Very high |\r\n\r\n**Key insight from Lead 5**: Teardown confirms Goertek MANUFACTURES the components but the core silicon and optical design are Apple IP. This is a classic ODM/JDM (joint design manufacturing) relationship, not a technology dependency where Goertek sources from Apple — rather, Apple designs, Goertek builds.\r\n\r\n---\r\n\r\n## Step 5: Competitive triangulation\r\n\r\nA's peers in precision manufacturing / Apple supply chain:\r\n\r\n| Peer | Company | Has VR optical patents? | Points to same pattern? |\r\n|------|---------|------------------------|------------------------|\r\n| C₁ | Luxshare (立讯精密) | Similar pattern — manufacturing patents, sparse core IP | Yes — also an Apple ODM |\r\n| C₂ | AAC Tech (瑞声科技) | Strong acoustic patents, weak optical | Yes |\r\n| C₃ | Foxconn (鸿海/工业富联) | Manufacturing patents only | Yes |\r\n| C₄ | Sunny Optical (舜宇光学) | **Different** — holds significant optical design patents | No — vertical integration |\r\n\r\n**Triangulation result**: 3/4 peers show the same pattern (ODM model), but Sunny Optical is different (vertically integrated). The pattern confirms that the \"patent gap\" is structural — these are contract manufacturers, not technology developers. The \"gap\" reflects their business model, not a supplier dependency.\r\n\r\n**Triangulation boost**: ≥3 peers same pattern → but the pattern points to a BUSINESS MODEL explanation (ODM, not technology sourcing), not a specific supplier B. **No boost applied.** Insight: this workflow correctly identified that the \"gap\" is a business-model artifact.\r\n\r\n---\r\n\r\n## Step 6: Fusion + confidence\r\n\r\n- Lead 5 (teardown) = Very high — confirms manufacturing relationship\r\n- Lead 3 (talent flow) = Medium — supports tech transfer direction (Apple → Goertek)\r\n- Leads 1 and 2 do NOT point to a supplier B because Goertek is itself the supplier (to Apple)\r\n- The workflow correctly identifies: **Goertek is NOT the one with the patent gap in the traditional sense. Goertek IS the supplier (B) in this analysis.**\r\n\r\n**Workflow adaptation**: In this case, the typical patent-gap→supplier direction doesn't apply. The more useful analysis would be the **reverse workflow**: who are Goertek's downstream customers?\r\n\r\n**Base confidence for forward analysis**: Cannot infer dependency (Goertek is the manufacturer, not the technology seeker).\r\n\r\n---\r\n\r\n## Step 7: Financial cross-validation\r\n\r\nSince Goertek IS the supplier to Apple (not the other way around), switch perspective:\r\n\r\n### Goertek as \"B\" (supplier):\r\n- **Goertek's customer concentration** (2023 annual report): Customer #1 = **46.7%** of revenue\r\n- This unnamed customer #1 is widely understood to be Apple\r\n- Customer #2 = 8.3%, Customer #3 = 5.1%\r\n- **Extreme single-customer concentration**: >40% → very high dependency risk\r\n\r\n**Financial corroboration**: **Strong**. The customer concentration data + teardown confirmation make the Apple→Goertek relationship unambiguous.\r\n\r\n---\r\n\r\n## Step 8: Alternative-hypothesis check\r\n\r\n| Hypothesis | Excluded? | Basis |\r\n|-----------|-----------|-------|\r\n| Goertek develops core VR optical IP in-house | Partially | Patents show assembly/testing IP, not core optical design |\r\n| Goertek is an NPE / licensor | Yes | Goertek is a manufacturer with physical facilities |\r\n| Apple-Goertek is purely financial | Yes | Physical teardown evidence + revenue concentration |\r\n| Goertek has a second major customer reducing concentration risk | Partially | Customer #2 at 8.3% is small but growing (possibly Meta) |\r\n\r\n---\r\n\r\n## Step 9: Counter-intelligence check\r\n\r\n| Check | Result |\r\n|-------|--------|\r\n| Patent ownership abnormally dispersed? | No |\r\n| Shell companies holding patents? | No |\r\n| Supplier disclosure vague? | **Yes** — Apple NDA prevents detailed disclosure |\r\n| Counter-intelligence risk | **Low** — NDA-driven opacity, not deliberate deception |\r\n\r\n---\r\n\r\n## Step 10: Report Summary\r\n\r\n### Key Finding\r\n\r\nThe patent-gap→supplier workflow, when applied to Goertek, revealed that **Goertek IS the supplier**, not the technology seeker. This is a case where the skill correctly identified that the standard forward workflow doesn't apply and pivoted to the more appropriate analysis:\r\n\r\n**Goertek is an ODM/JDM manufacturer for Apple's VR/audio products.** The patent \"gap\" in VR optics reflects Goertek's business model (contract manufacturing), not a dependency on an external technology supplier.\r\n\r\n### What We Actually Learned\r\n\r\n1. **The skill's self-awareness worked**: it correctly flagged that the \"gap\" is structural (ODM model), not a supply-chain dependency\r\n2. **Triangulation confirmed the pattern**: 3/4 peers show the same ODM business model\r\n3. **Financial data confirmed extreme customer concentration**: 46.7% from a single customer (Apple)\r\n4. **The correct next step** is the reverse workflow: \"Who else depends on Goertek's manufacturing capability?\" — which would identify Meta, Sony, and other VR/audio brands\r\n\r\n### Investment Implications\r\n\r\n- **Risk**: Extreme single-customer concentration (46.7%) → Goertek's fate is tied to Apple's product cycle\r\n- **Opportunity**: If Goertek diversifies into Meta/Sony VR manufacturing, revenue concentration risk decreases\r\n- **Follow-up**: Monitor Goertek's customer #2 revenue growth for diversification signals\r\n\r\n---\r\n\r\n## Lessons for the Skill\r\n\r\nThis case study demonstrates several v0.3.0 features in action:\r\n\r\n1. **The skill should recognize ODM/JDM patterns**: When a company's \"gap\" reflects its contract manufacturing business model, pivot to reverse workflow\r\n2. **Triangulation can reveal structural patterns**, not just supplier convergence\r\n3. **Financial cross-validation in this case confirmed extreme concentration** — a key risk signal\r\n4. **Industry context matters**: consumer electronics ODM players will ALWAYS show patent gaps in core technology; this is normal\n\nFile v0.1.2:examples/sample-output.md\n\n# Sample Output\r\n\r\nA worked example showing the full v0.4.0 report format for a **fictional** case with triangulation, negative verification, patent quality scoring, and evidence freshness tracking. All companies, patents, and data below are illustrative only.\r\n\r\n---\r\n\r\n## Input\r\n\r\n- **Direction**: Forward\r\n- **Target A**: NexRouter Inc. (fictional ticker: NXR)\r\n- **Product P**: NexRouter 5G Pro router\r\n- **Technology T**: 5G mmWave antenna technology\r\n- **Background**: NexRouter's 5G Pro uses mmWave antennas, but NexRouter's filings show no mmWave-antenna patents.\r\n\r\n## Step 0: Applicability\r\n\r\nHardware / networking equipment → **Strong applicability**. Proceed.\r\n\r\n## Step 1: Trigger + Direction\r\n\r\n- Direction: Forward\r\n- A = NexRouter Inc.\r\n- P = NexRouter 5G Pro\r\n- T = 5G mmWave antenna technology\r\n- Preliminary: NexRouter's main R&D is in router firmware / SoC, not RF antenna design. Gap likely.\r\n\r\n## Step 2: Patent gap confirmation\r\n\r\nSearched NexRouter + subsidiaries in patent DB (CN + US + PCT) for \"mmWave antenna\" / IPC H01Q:\r\n- NexRouter patents: 12 total, all in firmware / routing protocols / SoC. **Zero** mmWave-antenna patents.\r\n- Gap status: **Confirmed**.\r\n\r\nExcluded legitimate-use scenarios:\r\n- Cross-licensing: no public record found.\r\n- Acquisition: no mmWave-antenna company acquired.\r\n- Patent pool: mmWave antenna not part of Avanci / MPEG-LA pools.\r\n- Open-source: no open-source mmWave-antenna implementation applicable.\r\n\r\n## Step 3: Trade secret assessment\r\n\r\n| Indicator | Assessment | Signal |\r\n|-----------|------------|--------|\r\n| 1. Industry norms | mmWave antenna is RF hardware, not a process/formula; easy to reverse-engineer | Dependency |\r\n| 2. NDA culture | NexRouter is a standard networking company; no notable NDA litigation | Neutral |\r\n| 3. Patent filing pattern | NexRouter files actively in firmware/SoC but zero in RF — no strategic selectivity | Neutral |\r\n| 4. R&D vs. patent output | NexRouter lists 0 RF engineers and 0 RF R&D budget in filings | Dependency |\r\n| 5. Technology nature | mmWave antenna is a physical hardware component, identifiable in teardown | Dependency |\r\n| 6. Trade secret litigation | No trade secret litigation in mmWave antenna industry | Dependency |\r\n\r\n**Trade secret likelihood**: 0/6 → **Trade secret unlikely**. Proceed.\r\n\r\n## Step 4: Multi-lead parallel verification\r\n\r\n### Lead 1 · Patent ownership + quality\r\n\r\nWaveAnt Technologies holds 47 mmWave-antenna patents. **Patent quality scoring**:\r\n\r\n| Quality factor | Score | Basis |\r\n|----------------|-------|-------|\r\n| Claim breadth | 3/3 | Independent claims cover the phased-array steering mechanism itself |\r\n| Citation impact | 3/3 | 41% of later mmWave-antenna patents cite WaveAnt's core family |\r\n| Family size | 3/3 | PCT + US/CN/EU national entries |\r\n| Legal status | 2/3 | Granted, maintenance current on core family |\r\n| **Total** | **11/12** | **Core quality** |\r\n\r\nLead 1 confidence: **High** (Core quality + valid timeline, core filings 2019-2021, before 5G Pro launch 2024). Evidence date: 2024-08 (current).\r\n\r\n### Leads 2-5\r\n\r\n| Lead | Hit | Detail | Confidence | Evidence date | Freshness |\r\n|------|-----|--------|------------|---------------|-----------|\r\n| 2. Corporate affiliation | Yes | Co-founded JV \"NexWave RF\" (2023), 60/40 split | High | 2023-06 | Stale (>18mo) |\r\n| 3. Talent flow | Yes | 3 WaveAnt engineers → NexRouter (2023-24) | Medium | 2024-03 | Moderate |\r\n| 4. Procurement | Yes | 2023 annual report: \"mmWave antenna module\" top-5 procurement, supplier unnamed | Medium | 2024-04 | Moderate |\r\n| 5. Product teardown | Yes | TechInsights teardown identifies \"WaveAnt WN-5G39\" module | Very high | 2025-11 | **Current** |\r\n\r\n**Freshness note**: Lead 2 (JV) evidence is from 2023 — flagged as stale, down-weighted to Medium. Leads 4/5 are the freshest and strongest.\r\n\r\n## Step 5: Competitive triangulation\r\n\r\nA's peers in networking equipment:\r\n\r\n| Peer | Has mmWave patents? | Points to same B? | B's quality for peer? |\r\n|------|--------------------|-------------------|----------------------|\r\n| C₁ RouteMax | Zero | Yes — teardown shows WN-5G37 | Core (WaveAnt) |\r\n| C₂ NetFusion | Zero | Yes — procurement category matches | Core |\r\n| C₃ SpeedLinx | Zero | Yes — JV with WaveAnt (2022) | Core |\r\n| C₄ TeleCore | **Has own** | No — internal development | — |\r\n\r\n**Triangulation result**: 3/4 peers all lack T patents, all point to WaveAnt, and WaveAnt's patents score **Core** quality across the board. **Triangulation boost: +2 levels** (quality-confirmed convergence).\r\n\r\n## Step 6: Negative verification\r\n\r\nActively searched for evidence that WaveAnt is NOT NexRouter's supplier:\r\n\r\n| Negative check | Result |\r\n|----------------|--------|\r\n| Revenue mismatch | **No** — WaveAnt revenue (~¥900M) is consistent with supplying 5+ networking companies; NexRouter's estimated antenna spend (~¥380M) is a plausible share |\r\n| Product-category mismatch | **No** — WaveAnt's catalog includes the exact module type (5G mmWave array) found in teardown |\r\n| Geographic infeasibility | **No** — both based in Guangdong, logistics normal |\r\n| Customer-list exclusion | **Partially checked** — WaveAnt is pre-IPO, no public customer list; could not fully verify |\r\n| Contractor/competitor | **No** — WaveAnt makes antenna modules; NexRouter makes routers; not competitors |\r\n| Third-party licensing | **No** — teardown shows WaveAnt-branded module, not a third-party rebrand |\r\n\r\n**Negative verification outcome**: Search performed, no fundamental negative evidence found. One gap: customer list unavailable (WaveAnt pre-IPO). **Negative verification passed** (with noted data gap).\r\n\r\n## Step 7: Fusion + confidence\r\n\r\n- Leads: 5/5 hit. Lead 5 (teardown, current evidence) = very high. Lead 1 (Core quality) = high. Leads 2-4 = medium (with freshness adjustments).\r\n- Base confidence: **High** (≥2 high leads corroborating).\r\n- Triangulation boost: +2 → remains High (already at ceiling).\r\n- Negative verification: passed → no reduction.\r\n- Evidence freshness: strongest evidence (teardown 2025-11) is current → no freshness cap.\r\n- **Pre-financial confidence: High**.\r\n\r\nRole determination: Leads 4/5 hit → WaveAnt is a **direct supplier**.\r\n\r\n## Step 8: Financial cross-validation\r\n\r\n| Check | Result | Corroboration |\r\n|-------|--------|---------------|\r\n| NexRouter top-5 suppliers | Supplier #3 \"antenna module supplier\" = 12.4% of procurement, ¥380M — matches WaveAnt revenue profile | **Medium** |\r\n| WaveAnt customer concentration (pre-IPO filings) | Customer #1 = 22% of revenue, profile matches NexRouter | **Medium** |\r\n| AR/AP patterns | WaveAnt AR 60% from one customer, 90-day terms — matches NexRouter AP | **Medium** |\r\n\r\n**Financial corroboration**: **Medium** (category + revenue scale match, no explicit naming due to pre-IPO opacity). **No cap applied** (medium corroboration + strong physical evidence).\r\n\r\n## Step 9: Alternative-hypothesis check\r\n\r\n| Hypothesis | Excluded? | Basis |\r\n|-----------|-----------|-------|\r\n| Gap due to cross-licensing | Yes | No public record |\r\n| WaveAnt is NPE | Yes | Manufactures physical modules (teardown) |\r\n| Third-party module embedding WaveAnt patent | Yes | Teardown shows direct WaveAnt module |\r\n| Pure financial investment tie | Yes | JV operational + direct supply |\r\n| Trade secret protection | Yes | 0/6 indicators, no RF R&D |\r\n| Relationship economically immaterial | Yes | 12.4% of procurement |\r\n| B's patents are peripheral (paper tiger) | Yes | Core quality 11/12 |\r\n\r\nUnexcluded: none material.\r\n\r\n## Step 10: Counter-intelligence check\r\n\r\n| Check | Result |\r\n|-------|--------|\r\n| Patent ownership dispersed? | No — 47 patents under WaveAnt directly |\r\n| Shell companies? | No — real operating company |\r\n| Deliberate non-filing? | Unlikely — files aggressively in core areas |\r\n| Frequent transfers? | No |\r\n| Supplier disclosure vague? | **Flag** — annual report doesn't name supplier |\r\n\r\nCounter-intelligence risk: **Low**. One minor flag.\r\n\r\n## Step 11: Output report\r\n\r\n### Supply-Chain Inference Report — NexRouter Inc.\r\n\r\n**Direction**: Forward (patent gap → supplier)\r\n\r\n**Trade secret assessment**: Unlikely (0/6 indicators).\r\n\r\n**Patent quality**: WaveAnt's core family scores **11/12 (Core)** — broad claims, high citations, international family, active status.\r\n\r\n**Evidence freshness**: Teardown (2025-11) current; JV evidence (2023) flagged stale and down-weighted.\r\n\r\n**Triangulation**: 3/4 peers share the gap, all converge on WaveAnt, Core-quality confirmation. +2 boost.\r\n\r\n**Negative verification**: Performed, passed (with pre-IPO customer-list data gap noted).\r\n\r\n**Relationship determination**: WaveAnt Technologies is a **direct supplier** of mmWave antenna modules to NexRouter.\r\n\r\n**Evidence chain**:\r\n\r\n| # | Lead | Hit | Confidence | Quality/Freshness |\r\n|---|------|-----|------------|-------------------|\r\n| 1 | Patent ownership | WaveAnt 47 patents, Core 11/12 | High | Quality-adjusted |\r\n| 2 | Corporate affiliation | JV \"NexWave RF\" (2023) | Medium | Stale evidence, down-weighted |\r\n| 3 | Talent flow | 3 engineers → NexRouter | Medium | Moderate freshness |\r\n| 4 | Procurement | Top-5 category, unnamed | Medium | Moderate freshness |\r\n| 5 | Product teardown | \"WaveAnt WN-5G39\" module | Very high | **Current** (2025-11) |\r\n\r\n**Confidence stack**:\r\n- Base fusion: 5/5 leads, 2 high + very high corroborating → **High**\r\n- Triangulation: +2 (quality-confirmed) → High (ceiling)\r\n- Negative verification: passed → no change\r\n- Evidence freshness: no cap (strongest evidence current)\r\n- Financial: Medium corroboration → no cap\r\n- **Final confidence: High**\r\n\r\n**Alternative hypotheses**: All material alternatives excluded.\r\n\r\n**Counter-intelligence risk**: Low.\r\n\r\n**Follow-up verification**:\r\n- Re-verify JV \"NexWave RF\" activity (stale evidence, 2023)\r\n- Await WaveAnt IPO filings for explicit customer naming\r\n- Monitor WaveAnt patent filings post-2024 for co-development indicators\r\n- Watch RouteMax/NetFusion teardown results (triangulation peers)\n\nFile v0.1.2:skill-card.md\n\n## Description:\n\nInfers likely supplier, customer, or technology-licensing relationships from patent gaps through trade-secret checks, multi-lead verification, negative verification, patent quality scoring, evidence freshness tracking, and financial cross-validation.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[tianzhiceng297-boop](https://clawhub.ai/user/tianzhiceng297-boop)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nAnalysts and agents use this skill to research patent gaps, supplier or customer relationships, technology dependencies, and competitive supply-chain signals in hardware, manufacturing, pharma, materials, and related industries. It is designed to produce confidence-scored inference reports rather than definitive legal, procurement, or investment conclusions.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: Patent-gap and supply-chain inferences can be wrong or stale.\n\nMitigation: Review the skill's confidence ratings, evidence timestamps, negative verification results, and financial corroboration before relying on a conclusion.\n\nRisk: A patent gap can reflect trade-secret protection, normal component procurement, or an ODM business model rather than a supplier dependency.\n\nMitigation: Run the documented applicability check, trade-secret assessment, alternative-hypothesis checklist, and industry-specific guide before treating a patent gap as dependency evidence.\n\nRisk: The workflow may be misused as a general company research or investment-decision assistant.\n\nMitigation: Use it only for the specialized public-source competitive-intelligence workflow described by the release evidence, and keep final business or investment decisions under human review.\n\n## Reference(s):\n\n- [Detailed Workflow](references/workflow.md)\n- [Checklists and Confidence Rules](references/checklists.md)\n- [Industry-Specific Patent and Trade Secret Guide](references/industry-guide.md)\n- [Sample Output](examples/sample-output.md)\n- [Real Case Study](examples/real-case.md)\n- [Changelog](CHANGELOG.md)\n\n## Skill Output:\n\n**Output Type(s):** [text, markdown, guidance]\n\n**Output Format:** [Markdown supply-chain inference report with confidence tables and follow-up recommendations]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [Includes evidence timestamps, confidence levels, patent quality scores, negative verification results, financial corroboration, and alternative-hypothesis checks.]\n\n## Skill Version(s):\n\n0.1.2 (source: ClawHub release metadata; artifact frontmatter and changelog state 0.4.0)\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 v0.1.1: 11 files, 33310 bytes\n\nFiles: CHANGELOG.md (4966b), examples (0b), examples/real-case.md (9886b), examples/sample-output.md (7894b), README.md (8330b), references (0b), references/checklists.md (9857b), references/industry-guide.md (9161b), references/workflow.md (14594b), SKILL.md (12742b), _meta.json (142b)\n\nFile v0.1.1:SKILL.md\n\n---\nname: patent-gap-supply-chain\ndescription: 'Competitive intelligence workflow that infers supply-chain\n  relationships from patent gaps — with trade secret assessment, competitive\n  triangulation, reverse workflow (supplier→client), and financial cross-validation.\n  通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、反向工作流\n  （供应商→客户）、财务数据交叉验证于一体。Use when researching\n  tech/hardware/manufacturing/pharma stocks, analyzing industry supply chains,\n  identifying suppliers or customers, investigating patent gaps, or inferring\n  supply-chain relationships from patent ownership. Supports both forward\n  (patent gap → supplier) and reverse (technology owner → downstream clients)\n  analysis directions. 触发场景：科技股研究、行业分析、产业链推断、供应商识别、客户挖掘、\n  专利缺口分析、专利归属查询、技术依赖分析、上下游推断 / Triggers:\n  patent gap analysis, supply chain inference, supplier identification,\n  customer identification, technology dependency, competitive intelligence,\n  tech stock research, industry analysis, trade secret assessment, financial\n  cross-validation, reverse supply-chain analysis. The skill starts from a\n  patent gap (or known technology owner), evaluates trade secret likelihood,\n  activates five verification leads with competitive triangulation, then\n  cross-validates against financial disclosures, outputting a\n  confidence-scored supply-chain inference report.'\nversion: 0.3.0\nagent_created: true\ndisable: true\n---\n\n# Patent Gap Supply Chain Inference\n\n## Overview\n\nThis skill implements a competitive intelligence workflow: \"patent gap → supply chain inference.\" Core insight: when target company A's product uses technology T, but A holds no patents related to T, there may be a technology-licensing or supply-chain dependency. By searching T's patent ownership to identify company B, then activating multiple verification leads, one can infer whether B has entered A's supply chain.\n\n**Critical caveat**: A patent gap can also mean A protects T as a **trade secret / know-how** rather than through patents. Patents require public disclosure; many companies (especially in semiconductor processes, chemical formulations, pharma manufacturing) deliberately choose trade secrets. This skill explicitly evaluates this distinction before concluding supply-chain dependency.\n\n**v0.3.0 new capabilities**:\n- **Competitive triangulation**: If A's peers C₁, C₂...Cₙ also lack T's patents and all point to B, inference confidence gets a significant boost. Cross-company corroboration is one of the strongest signals.\n- **Reverse workflow** (supplier → clients): Start from technology owner B (who holds monopoly patents in T) and identify downstream companies A₁, A₂...Aₙ that likely depend on B's technology. This is the more common investment research use case.\n- **Financial cross-validation**: After inferring a supply-chain relationship, cross-validate against A's top-supplier disclosures, B's customer concentration data, and accounts receivable patterns. Inference without financial corroboration is capped at \"medium\" confidence.\n- **Industry-specific guidance**: See `references/industry-guide.md` for semiconductor, pharma, new energy, consumer electronics, and materials verticals — each has different patent/trade-secret dynamics.\n\nThe skill downgrades \"patents\" from the sole evidence chain to an entry lead, enforcing multi-lead parallel verification, competitive triangulation, financial cross-validation, alternative-hypothesis checks, and counter-intelligence checks to prevent overconfident wrong conclusions.\n\n## When to Use\n\nTrigger when any of the following applies:\n\n**Forward direction** (patent gap → find supplier):\n- Researching tech, hardware, manufacturing, pharma, or materials companies where supply chain matters\n- Discovering or suspecting a company uses a technology but lacks related patents\n- Explicitly asked to do patent ownership analysis, technology dependency analysis, or supply-chain inference\n\n**Reverse direction** (technology owner → find clients) (NEW v0.3.0):\n- A company B holds dominant/monopoly patents in a technology T and you want to identify B's downstream customers\n- Investment thesis: \"B is the sole/dominant supplier of T; companies using T must be B's customers\"\n- Asked to identify companies that depend on a specific technology or component\n\nTrigger keywords (bilingual): 科技股研究, 行业分析, 产业链, 供应链, 供应商识别, 客户挖掘, 上下游推断, 专利缺口, 专利归属, 技术依赖, patent gap, supply chain, supplier identification, customer identification, technology dependency, competitive intelligence, reverse supply chain.\n\n## Step 0: Applicability Pre-check (mandatory)\n\nBefore running the workflow, judge whether the target type is applicable. See `references/industry-guide.md` for industry-specific dynamics:\n\n| Target type | Applicability | Handling |\n|-------------|---------------|----------|\n| Hardware / manufacturing / pharma / materials | Strong | Proceed. Reference `industry-guide.md` for vertical-specific signals |\n| Software / internet | Weak | Warn that patent protection is weak and tech is hidden; suggest code / open-source intelligence; may proceed at reduced confidence |\n| Business-model innovation | Not applicable | Stop; suggest financial reports / prospectus / business-model analysis instead |\n\nWhen not applicable, clearly inform the user and suggest alternatives. Do not force execution.\n\n## Direction Selection (NEW v0.3.0)\n\nBefore proceeding, determine the analysis direction:\n\n| Direction | Starting point | Goal | Trigger phrase examples |\n|-----------|---------------|------|------------------------|\n| **Forward** | Company A uses T but lacks T's patents | Find who supplies T to A | \"Identify XX's suppliers\", \"Who provides XX's technology?\" |\n| **Reverse** | Company B holds dominant T patents | Find who depends on B's T | \"Who are B's customers?\", \"What downstream companies need this patent?\" |\n\nBoth directions share the same verification leads, but the entry point and initial search differ. See `references/workflow.md` for detailed reverse-workflow operations.\n\n## Workflow Overview\n\nThe full workflow has 10 steps. Detailed operations in `references/workflow.md`:\n\n1. **Trigger lead identification**: Confirm target company, product, technology; determine forward or reverse direction\n2. **Patent gap confirmation** (forward) / **Patent dominance confirmation** (reverse): Search patent portfolio, determine gap validity or market dominance\n3. **Trade secret assessment**: Evaluate whether gap reflects deliberate trade secret / know-how protection\n4. **Multi-lead parallel verification**: Activate 5 leads — patent ownership, corporate affiliation, talent flow, procurement, product teardown\n5. **Competitive triangulation** (NEW v0.3.0): Check A's peers/competitors for the same patent gap. If ≥2 peers also lack T's patents and all point to B, boost confidence.\n6. **Multi-lead fusion + confidence scoring**: ≥2 high-confidence leads corroborating required; triangulation provides confidence multiplier\n7. **Financial cross-validation** (NEW v0.3.0): Cross-validate inferred relationship against A's supplier disclosures / B's customer concentration / accounts receivable. Inference without financial corroboration is capped.\n8. **Alternative-hypothesis check**: Explicitly exclude alternative explanations; see `references/checklists.md`\n9. **Counter-intelligence check**: Check for shell companies / dispersed patents; see `references/checklists.md`\n10. **Output report**: Relationship determination + evidence chain + triangulation result + financial corroboration + confidence + alternative hypotheses + trade secret assessment + counter-intel risk + follow-up verification\n\n## Input/Output Contract\n\n**Input**:\n- Target company (A for forward, B for reverse) — name / ticker\n- Technology T or product P of interest (optional; may be identified during analysis)\n- Analysis direction: forward or reverse (default: forward if user mentions \"supplier\", reverse if user mentions \"customer\" or \"downstream\")\n- Known background info (optional)\n\n**Output**: Supply-chain inference report, including:\n- **Direction**: forward or reverse analysis\n- **Relationship determination**: whether a supply-chain relationship exists + role (direct supplier / technology licensor / customer / indirect affiliation / NPE license-only)\n- **Trade secret assessment**: likelihood + indicator breakdown\n- **Evidence chain**: per-lead hit status + per-lead confidence + corroboration\n- **Triangulation result** (v0.3.0): which peers confirmed the same gap + triangulation boost applied\n- **Financial corroboration** (v0.3.0): supplier disclosure hits, customer concentration data, AR patterns, corroboration status\n- **Overall confidence**: high / medium / low + rationale (including triangulation boost and financial cap if applicable)\n- **Alternative-hypothesis list**: excluded + unexcluded\n- **Counter-intelligence risk flag**: shell companies / dispersed patents\n- **Follow-up verification suggestions**: next evidence to gather\n\n## Tool Integration\n\nAvailable tools per lead (by availability priority):\n\n| Lead | Tools | Purpose |\n|------|-------|---------|\n| Patent search | tushare (patent API), PatSnap patsnap-search connector | Search A/B/peers patents, T's patent ownership |\n| Corporate affiliation | Tianyancha tyc-mcp, Qichacha qcc-company, Qixinbao qixinbao-mcp | A-B investment / JV / executive overlap |\n| Listed-co data | tushare, westock-mcp, westock-tool | Financial report supplier disclosure, prospectus |\n| Talent flow | WebSearch, job-site search | B→A technical hires |\n| Procurement | WebSearch, gov-procurement portals, tender announcements | A's tender / supplier announcements |\n| Product teardown | WebSearch, professional teardown reports | A's product BOM / teardown for B components |\n| Trade secret assessment | WebSearch, legal databases, company reports | NDAs, non-competes, trade secret litigation, R&D disclosures |\n| Competitive triangulation (v0.3.0) | tushare, WebSearch, industry reports | A's peer patent portfolios, cross-company gap confirmation |\n| Financial cross-validation (v0.3.0) | westock-mcp, tushare, annual report filings | Supplier concentration, customer concentration, AR aging, related-party transactions |\n| Industry guidance (v0.3.0) | `references/industry-guide.md` | Vertical-specific patent/trade-secret dynamics and verification strategies |\n\nWhen a tool is unavailable, fall back to WebSearch for public info and reduce that lead's confidence accordingly.\n\n## Key Principles\n\n- **Patents are an entry lead, not final evidence**: A patent gap triggers verification but does not directly conclude\n- **Trade secrets must be ruled out before concluding dependency**: A patent gap may reflect deliberate trade secret / know-how protection\n- **Procurement/teardown leads override trade secret hypothesis**: Physical evidence trumps trade secret likelihood\n- **Competitive triangulation strengthens inference** (v0.3.0): Cross-company corroboration is one of the strongest signals — a gap shared by multiple peers pointing to the same supplier is much harder to dismiss\n- **Financial cross-validation is required for high-confidence conclusions** (v0.3.0): No supply-chain inference reaches \"high\" without at least one form of financial corroboration (supplier disclosure, customer concentration, or AR evidence)\n- **Multi-lead corroboration**: ≥2 high-confidence leads corroborating required to conclude\n- **Explicit confidence scoring**: Every lead and inference gets a confidence score (high / medium / low)\n- **Alternative hypotheses are mandatory**: Every inference must explicitly exclude alternative explanations\n- **Counter-intelligence check is mandatory**: Watch for shell companies / dispersed patents hiding the real supply chain\n\n## Resources\n\n### references/\n- `references/workflow.md`: Detailed 10-step operations + reverse workflow section\n- `references/checklists.md`: All checklists — alternative hypotheses, trade secret indicators, triangulation, financial cross-validation, counter-intelligence, confidence scoring rules\n- `references/industry-guide.md` (NEW v0.3.0): Vertical-specific guidance for semiconductor, pharma, new energy, consumer electronics, and materials\n\n### examples/\n- `examples/sample-output.md`: Fictional case showing the full 10-step report format\n- `examples/real-case.md` (NEW v0.3.0): A real Chinese A-share case study with actual company names and data\n\nFile v0.1.1:README.md\n\n# patent-gap-supply-chain\n\n通过\"专利缺口\"反推产业链关系的竞争情报工作流 skill，适用于 WorkBuddy / Codebuddy / Claude Code / Codex。\n\nA competitive intelligence workflow skill that infers supply-chain relationships from patent gaps. Works with WorkBuddy / Codebuddy / Claude Code / Codex.\n\n---\n\n## 中文说明\n\n### 这是什么\n\n当研究科技股 / 硬件 / 制造 / 医药类公司时，发现标的 A 的产品使用了技术 T，但 A 没有 T 的相关专利——这套 skill 引导从专利缺口切入，**先判断缺口是真实外部依赖还是商业秘密/know-how保护**，再激活五条线索并行验证，竞对三角验证增强信号，财务数据交叉验证定论，输出带置信度的产业链关系推断报告。\n\n**v0.3.0 新增**：\n- **竞对三角验证**：A 的同行如果也缺相同专利、也指向同一家供应商 → 置信度大幅提升\n- **反向工作流**（供应商→客户）：从技术垄断者出发，找下游依赖客户\n- **财务数据交叉验证**：供应商/客户集中度、应收账款、关联交易 — 没有财报佐证的推断置信度封顶\n- **行业特化参考**：半导体/制药/新能源/消费电子/材料五大行业指南\n- **真实案例**：歌尔股份 & Apple VR 供应链完整分析\n\n**v0.2.0**：商业秘密/know-how 评估（6 项指标）\n\n### 核心设计\n\n- **商业秘密先排除再定论**：专利缺口可能是商业秘密保护策略，必须先评估\n- **竞对三角验证** (v0.3.0)：多同行同一缺口 → 同一供应商，是最强信号之一\n- **财务交叉验证** (v0.3.0)：无财报佐证的推断置信度最多\"中等\"，需供应商披露/客户集中度/应收账款等硬数据\n- **双向工作流** (v0.3.0)：正向（缺口→供应商）+ 反向（专利所有者→下游客户）\n- **专利降级为入口线索**：专利缺口只触发验证，不直接定论\n- **多线索交叉印证**：≥2 条高置信线索相互印证方可定论\n- **替代假设必查**：显式排除交叉授权 / NPE / 第三方模组 / 商业秘密等替代解释\n- **反情报对抗必查**：警惕壳公司 / 分散专利掩盖真实供应链\n\n### 安装\n\n    git clone https://github.com/tianzhiceng297-boop/patent-gap-supply-chain.git ~/.workbuddy/skills/patent-gap-supply-chain\n\n或手动复制 `SKILL.md` 和 `references/` 到 `~/.workbuddy/skills/patent-gap-supply-chain/`。\n\n### 兼容性 / Compatibility\n\n| Agent | 加载方式 |\n|------|---------|\n| WorkBuddy / Codebuddy | 放入 `~/.workbuddy/skills/`，自动加载 + 触发词触发（原生支持） |\n| Claude Code | 将 `SKILL.md` 内容追加到项目 `CLAUDE.md`，或放入 `.claude/commands/` 作为自定义命令；`references/` 按需在命令中引用 |\n| Codex (OpenAI) | 将 `SKILL.md` 内容追加到项目 `AGENTS.md`；`references/` 按需引用 |\n| 其他 agent | 将 `SKILL.md` 作为 system prompt 或项目指令注入；`references/` 按需提供 |\n\n> 本 skill 的指令内容是通用 markdown，任何能读取并执行 markdown 指令的 agent 均可使用。区别仅在于加载机制：WorkBuddy 原生支持自动触发，其他 agent 需手动注入或配置。\n\n### 触发场景\n\n- \"研究 XX 科技股\"\n- \"分析 XX 公司的供应链 / 供应商\"\n- \"查一下 XX 技术的专利归属\"\n- \"XX 公司用了某技术但好像没专利\"\n- \"识别 XX 的供应商\"\n- \"XX 公司的下游客户有哪些？\" (反向工作流)\n- \"哪些公司依赖 XX 的技术？\" (反向工作流)\n\n### 适用边界\n\n| 标的类型 | 适用性 |\n|---------|--------|\n| 硬件 / 制造 / 医药 / 材料 | 强适用 |\n| 软件 / 互联网 | 弱适用（降低置信度） |\n| 商业模式创新 | 不适用 |\n\n---\n\n## English\n\n### What it is\n\nWhen researching tech / hardware / manufacturing / pharma companies, you find that target A's product uses technology T, but A holds no patents related to T. This skill starts from the patent gap, **first evaluates whether the gap reflects true external dependency or deliberate trade secret / know-how protection**, then activates five parallel verification leads, competitive triangulation, and financial cross-validation, outputting a confidence-scored supply-chain inference report.\n\n**New in v0.3.0**:\n- **Competitive triangulation**: Cross-check A's peers for the same patent gap — multi-company corroboration is one of the strongest signals\n- **Reverse workflow** (supplier → clients): Start from a technology monopoly holder and identify downstream customers\n- **Financial cross-validation**: Supplier/customer concentration, AR data, related-party disclosures — no inference reaches \"high\" without financial corroboration\n- **Industry-specific guide**: Semiconductor, pharma, new energy, consumer electronics, materials verticals\n- **Real case study**: Goertek (歌尔股份) & Apple VR supply chain\n\n**v0.2.0**: Trade secret / know-how assessment (6 indicators)\n\n### Core design\n\n- **Trade secrets ruled out before concluding**: patent gap may reflect trade secret strategy — must assess first\n- **Competitive triangulation** (v0.3.0): multi-peer same gap → same supplier is a top-tier signal\n- **Financial cross-validation required** (v0.3.0): no \"high\" confidence without financial corroboration\n- **Bidirectional workflow** (v0.3.0): forward (gap → supplier) + reverse (patent owner → downstream clients)\n- **Patents demoted to an entry lead**: a patent gap triggers verification but does not directly conclude\n- **Multi-lead corroboration**: ≥2 high-confidence leads corroborating required to conclude\n- **Alternative hypotheses are mandatory**: explicitly exclude all alternative explanations\n- **Counter-intelligence check is mandatory**: watch for shell companies / dispersed patents\n\n### Installation\n\n    git clone https://github.com/tianzhiceng297-boop/patent-gap-supply-chain.git ~/.workbuddy/skills/patent-gap-supply-chain\n\nOr manually copy `SKILL.md` and `references/` to `~/.workbuddy/skills/patent-gap-supply-chain/`.\n\n### Compatibility\n\n| Agent | How to load |\n|-------|-------------|\n| WorkBuddy / Codebuddy | Drop into `~/.workbuddy/skills/`; auto-loaded + trigger-keyword activation (native) |\n| Claude Code | Append `SKILL.md` content to the project `CLAUDE.md`, or place under `.claude/commands/` as a custom command; reference `references/` as needed |\n| Codex (OpenAI) | Append `SKILL.md` content to the project `AGENTS.md`; reference `references/` as needed |\n| Other agents | Inject `SKILL.md` as a system prompt or project instruction; provide `references/` on demand |\n\n> The skill's instructions are plain markdown — any agent that can read and execute markdown instructions can use it. The only difference is the loading mechanism: WorkBuddy supports auto-triggering natively; other agents require manual injection or configuration.\n\n### Trigger scenarios\n\n- \"Research [tech stock]\"\n- \"Analyze [company]'s supply chain / suppliers\"\n- \"Look up patent ownership of [technology]\"\n- \"[Company] uses a technology but seems to have no patents\"\n- \"Identify [company]'s suppliers\"\n- \"Who are [company]'s downstream customers?\" (reverse workflow)\n- \"Which companies depend on [technology]?\" (reverse workflow)\n\n### Applicability\n\n| Target type | Applicability |\n|-------------|---------------|\n| Hardware / manufacturing / pharma / materials | Strong |\n| Software / internet | Weak (reduce confidence) |\n| Business-model innovation | Not applicable |\n\n---\n\n## File structure\n\n- `SKILL.md` — core: triggers, direction selection, 10-step workflow (triangulation + financial cross-validation), I/O contract, tool integration, key principles\n- `references/workflow.md` — detailed 10-step operations + reverse workflow appendix\n- `references/checklists.md` — alternative hypotheses, trade secret indicators, triangulation checklist, financial cross-validation checklist, counter-intelligence, confidence scoring rules with 3-layer adjustment\n- `references/industry-guide.md` (v0.3.0) — vertical-specific guidance: semiconductor, pharma, new energy, consumer electronics, materials/chemicals\n- `examples/sample-output.md` — fictional case showing full 10-step report format\n- `examples/real-case.md` (v0.3.0) — real A-share case: Goertek & Apple VR supply chain\n- `CHANGELOG.md` — version history\n\n## License\n\nMIT\n\nFile v0.1.1:_meta.json\n\n{\n  \"ownerId\": \"kn718c4g4psbwaq7bt8ft547cd8827z6\",\n  \"slug\": \"patent-gap-supply-chain\",\n  \"version\": \"0.1.1\",\n  \"publishedAt\": 1784612642274\n}\n\nFile v0.1.1:references/checklists.md\n\n# Checklists and Confidence Rules\r\n\r\n## Alternative-Hypothesis Checklist\r\n\r\nFor each inference, check each item and explicitly exclude alternative explanations.\r\n\r\n### Alternative explanations for the patent gap\r\n- [ ] Does A's patent gap stem from a cross-licensing agreement? (check announcements / news)\r\n- [ ] Did A inherit T usage rights via an acquisition? (check M&A history)\r\n- [ ] Does T belong to a patent pool? (e.g., Avanci, MPEG-LA, Via Licensing)\r\n- [ ] Is there an open-source implementation of T? (check open-source projects)\r\n- [ ] Does A hold T patents via a subsidiary / affiliate? (check related-entity patents)\r\n- [ ] Does A deliberately protect T as a trade secret / know-how rather than patenting it? (run Trade Secret Indicators checklist)\r\n- [ ] **v0.3.0**: Is the patent gap simply because T is a purchased component that A has no reason to patent? (common in consumer electronics — weak signal)\r\n\r\n### Alternative explanations for \"B entered the supply chain\"\r\n- [ ] Is B an NPE (non-practicing entity)? If so, B licenses only and is not in the supply chain\r\n- [ ] Does A use B's patent indirectly via a third-party module? The third party is the supplier; B is its upstream\r\n- [ ] Is B only a patent licensor with no physical transaction with A?\r\n- [ ] Is the A-B tie purely financial investment, not business cooperation?\r\n- [ ] **v0.3.0**: Is the A-B relationship economically immaterial? (financial cross-validation required to exclude)\r\n\r\n## Competitive Triangulation Checklist (NEW v0.3.0)\r\n\r\nFor each inference involving company A's patent gap, check against peers:\r\n\r\n### Peer identification\r\n- [ ] Identified 3-10 direct competitors / industry peers of A\r\n- [ ] Peer selection based on: same industry classification + similar product + similar revenue size\r\n- [ ] Both domestic and international peers considered where relevant\r\n\r\n### Gap verification\r\n- [ ] Peer C₁ patent portfolio checked for T → result: [has / partial / none]\r\n- [ ] Peer C₂ patent portfolio checked for T → result: [has / partial / none]\r\n- [ ] Peer Cₙ patent portfolio checked for T → result: [has / partial / none]\r\n\r\n### Convergence check\r\n- [ ] For peers with gaps, does T's ownership converge to the same B?\r\n- [ ] Any peers pointing to different B₂, B₃? → flag fragmented supply\r\n- [ ] Any peers with T patents? → A is the outlier, investigate why\r\n\r\n### Triangulation boost\r\n- [ ] ≥3 peers same gap → same B: **+2 confidence levels**\r\n- [ ] 2 peers same gap → same B: **+1 confidence level**\r\n- [ ] 1 peer same gap: note in report, no boost\r\n- [ ] Peers diverge: no boost, flag fragmentation\r\n- [ ] Triangulation cannot override trade secret cap or financial cross-validation cap\r\n\r\n## Financial Cross-Validation Checklist (NEW v0.3.0)\r\n\r\n### Forward direction\r\n- [ ] Checked A's top-5 supplier list in annual report\r\n- [ ] Checked A's supplier concentration ratio (>30% = single-supplier risk)\r\n- [ ] Matched A's procurement categories against B's product lines\r\n- [ ] If B is listed: checked B's top-5 customer list for A-like profile\r\n- [ ] Checked B's customer concentration (>20% from single customer)\r\n- [ ] Checked accounts receivable / payable patterns for correlation\r\n- [ ] Checked related-party transaction disclosures\r\n- [ ] Checked A's procurement growth vs. B's revenue growth (correlation)\r\n\r\n### Reverse direction\r\n- [ ] Checked B's top-5 customer list\r\n- [ ] Checked B's revenue by industry segment\r\n- [ ] For each candidate A: checked A's supplier list for B\r\n- [ ] For each candidate A: checked A's related-party disclosures\r\n- [ ] Estimated revenue contribution of each candidate (where data allows)\r\n\r\n### Financial corroboration scoring\r\n- [ ] **Strong corroboration**: B named as A's supplier OR A named as B's customer in official filings\r\n- [ ] **Medium corroboration**: Procurement category + revenue scale matches\r\n- [ ] **No corroboration**: No financial data found → **cap overall confidence at \"medium\"**\r\n- [ ] **Negative corroboration**: Financial data contradicts inference → **reduce confidence by 1 level**\r\n\r\n## Counter-Intelligence Checklist\r\n\r\n### Patent-ownership side\r\n- [ ] Is B's patent ownership abnormally dispersed? (multiple shell companies / affiliates holding separately)\r\n- [ ] Is there a shell company set up specifically to hold patents? (low registered capital, no real operations, abnormal address)\r\n- [ ] Are patent transfers frequent? (possibly to obscure true ownership)\r\n\r\n### A's patent-strategy side\r\n- [ ] Does A deliberately not file patents to hide its tech roadmap?\r\n- [ ] Do A's patent filings mismatch its actual product technology? (decoy filings)\r\n\r\n### Anomaly signals\r\n- [ ] Is A's supplier-disclosure abnormally terse or deliberately vague?\r\n- [ ] Are key technical staff employed via labor dispatch / outsourcing rather than directly?\r\n- [ ] Are there related entities with frequent changes of company name / legal representative / shareholders?\r\n\r\nIf any anomaly exists → flag \"counter-intelligence risk\" in the report.\r\n\r\n## Trade Secret / Know-How Indicators\r\n\r\n### Indicator 1 · Industry norms\r\n- [ ] T is in a trade-secret-dominant industry: semiconductor process, chemical formulation, pharma manufacturing process, food/beverage formula, metallurgy, specialty coatings, advanced materials synthesis\r\n- [ ] Industry is known for \"black box\" secrecy (e.g., TSMC process, ASML optics, DuPont formulas)\r\n\r\n### Indicator 2 · Company NDA / confidentiality culture\r\n- [ ] A has public record of aggressive NDA enforcement or trade secret litigation\r\n- [ ] A's employee contracts include strict non-compete + confidentiality clauses\r\n- [ ] A's facilities known for tight access control / \"clean room\" culture\r\n- [ ] A's technical publications / conference presentations notably sparse or redacted\r\n\r\n### Indicator 3 · Patent filing pattern (strategic selectivity)\r\n- [ ] A files patents across many domains but conspicuously avoids filing in T\r\n- [ ] A's patent portfolio shows carefully scoped filings (protecting what can be reverse-engineered, hiding what cannot)\r\n- [ ] A's patent filings in T's general category exist but intentionally omit key implementation details\r\n\r\n### Indicator 4 · R&D intensity vs. patent output\r\n- [ ] A has significant R&D team / budget in T's domain (check job postings, R&D headcount, R&D expense)\r\n- [ ] A produces few or zero patents in T despite the R&D investment\r\n- [ ] A's R&D-to-patent ratio in T is significantly lower than industry peers\r\n\r\n### Indicator 5 · Technology nature (protectability)\r\n- [ ] T is a process / method / formula / manufacturing technique (hard to reverse-engineer → good for trade secret)\r\n- [ ] T is a software algorithm embedded in a black-box system (source code hidden → trade secret feasible)\r\n- [ ] T would be difficult to prove infringement even if patented (→ low incentive to patent)\r\n\r\n### Indicator 6 · Trade secret litigation history\r\n- [ ] A has sued others for trade secret theft in T's domain\r\n- [ ] A's peers in the industry have active trade secret litigation\r\n- [ ] Industry news reports trade secret disputes related to T\r\n\r\n### Scoring\r\n- **4-6 indicators checked → High likelihood trade secret**: Patent gap likely strategic; default to \"no dependency\" unless procurement/teardown overrides\r\n- **2-3 indicators checked → Possible trade secret**: Record as alternative hypothesis; proceed normally\r\n- **0-1 indicators checked → Trade secret unlikely**: Gap likely reflects real dependency; proceed normally\r\n\r\n### Override signal\r\n- [ ] Lead 4 (procurement) returns high-confidence evidence of external sourcing → overrides trade secret hypothesis\r\n- [ ] Lead 5 (teardown) physically identifies third-party component → overrides trade secret hypothesis\r\n\r\n## Confidence Scoring Rules\r\n\r\n### Per-lead confidence\r\n\r\n| Lead | High | Medium | Low |\r\n|------|------|--------|-----|\r\n| Patent ownership | B holds T's core patents with valid timeline | B holds T-related patents but timeline questionable | T patents found but owner unclear |\r\n| Corporate affiliation | JV / mutual investment | Executive overlap | Same address / same legal rep (weak) |\r\n| Talent flow | Multiple B technical staff joined A | Single key person joined | Job postings only mention B background |\r\n| Procurement | B directly named as supplier | B's category mentioned | Category-related but no clear pointer |\r\n| Product teardown | Physically confirmed B component | Teardown report mentions but unconfirmed | No teardown report |\r\n\r\n### Overall confidence (base)\r\n\r\n- **High**: ≥2 high-confidence leads corroborating\r\n- **Medium**: 1 high + 1 medium corroborating, or multiple medium leads partially corroborating\r\n- **Low**: only 1 lead hit, or leads contradict each other\r\n- **Cannot infer**: 0 leads hit\r\n\r\n### Adjustment layers (apply in order, each CAN cap or boost):\r\n\r\n**Layer 1 — Trade secret adjustment** (v0.2.0):\r\n- Trade secret \"high\" + no procurement/teardown → cap at \"low\"; shift conclusion to \"likely internal trade secret\"\r\n- Trade secret \"high\" + procurement/teardown confirms external sourcing → no cap\r\n- Trade secret \"possible\" → no cap, record as alternative hypothesis\r\n\r\n**Layer 2 — Triangulation boost** (v0.3.0):\r\n- ≥3 peers same gap → same B: **boost +2 levels** (low→high, medium→high)\r\n- 2 peers same gap → same B: **boost +1 level**\r\n- Cannot override Layer 1 trade secret cap\r\n\r\n**Layer 3 — Financial cross-validation cap** (v0.3.0):\r\n- Strong financial corroboration → no cap\r\n- Medium financial corroboration → no cap\r\n- No financial corroboration → **cap at \"medium\"** (exception: procurement/teardown physical evidence can maintain \"high\" without financial data)\r\n- Negative corroboration → **reduce by 1 level**\r\n\r\n### Final confidence = min(base_confidence ± adjustments, Layer 3 cap)\n\nFile v0.1.1:references/industry-guide.md\n\n# Industry-Specific Patent & Trade Secret Guide\r\n\r\nDifferent industries have fundamentally different patent/trade-secret dynamics. This guide provides vertical-specific signals, verification strategies, and common pitfalls.\r\n\r\n---\r\n\r\n## 1. Semiconductor\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Manufacturing processes** (lithography, etching, deposition, packaging): Overwhelmingly trade secrets. TSMC, Samsung, Intel do NOT patent their process recipes. The 6-indicator trade secret assessment will almost always score 4+.\r\n- **Chip architectures / designs** (ISA, microarchitecture, interconnects): Strong patent coverage. ARM, x86, RISC-V are heavily patented.\r\n- **EDA tools / design methodologies**: Mixed. Some patented, some trade-secret algorithms.\r\n\r\n### Key Signals\r\n- A foundry's process node advancement without corresponding patents → trade secret, not gap\r\n- Fabless companies (e.g., MediaTek, UNISOC) lacking process patents → expected, they don't manufacture\r\n- Equipment makers (e.g., ASML, AMAT, LAM) → strong patent portfolios, less trade secret reliance\r\n\r\n### Verification Strategy\r\n- **Teardown > everything else**: Chip teardowns (TechInsights, ICWorks) are the gold standard\r\n- **Patent gap in chip design**: meaningful signal for IP licensing (e.g., ARM architecture license)\r\n- **Patent gap in manufacturing**: almost always trade secret, do not infer supplier dependency\r\n- **Financial cross-validation**: check foundry customer concentration in annual reports\r\n\r\n### Common A-Share Examples\r\n- SMIC (中芯国际): Process patents are sparse vs. TSMC; this reflects technology gap + trade secret strategy, not external sourcing\r\n- Hygon (海光信息): x86 license from AMD → genuine IP dependency\r\n- Cambricon (寒武纪): AI chip architecture patents → strong internal IP\r\n\r\n---\r\n\r\n## 2. Pharmaceuticals\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Compound patents** (small molecule structures): Heavily patented. The molecule IS the product.\r\n- **Manufacturing processes** (synthesis routes, purification, formulation): Often trade secrets. Same drug, different manufacturers may use different routes.\r\n- **Clinical trial data**: Regulatory exclusivity + trade secrets, not patents.\r\n- **Biologics / Biosimilars**: Complex mix. Cell line development and culture conditions are trade secrets.\r\n\r\n### Key Signals\r\n- Drug A has expired compound patent but no generic versions → manufacturing process is the barrier (trade secret or new process patent)\r\n- Biologic has zero process patents → expected, cell culture conditions are trade secrets\r\n- ANDA (generic drug application) filings → reveals which companies are trying to enter\r\n\r\n### Verification Strategy\r\n- **Patent expiry + generic entry timing**: The most reliable signal\r\n- **DMF (Drug Master File) filings**: Companies filing DMFs for an API are the actual suppliers\r\n- **CDE (药审中心) registration data**: China-specific regulatory filings\r\n- **Procurement/tender data**: Hospital procurement records, centralized drug procurement (集采)\r\n\r\n### Common A-Share Examples\r\n- Hengrui (恒瑞医药): Strong compound + formulation patents → internal R&D driven\r\n- WuXi AppTec (药明康德): CDMO, process patents + trade secrets → client dependency analysis\r\n- Generic manufacturers (Huahai 华海, Hisun 海正): Watch ANDA filing pipeline for target identification\r\n\r\n---\r\n\r\n## 3. New Energy / Battery\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Battery chemistry** (cathode materials, electrolyte formulations): Strong patent coverage. CATL, LG Energy, Panasonic all file aggressively.\r\n- **Manufacturing process** (coating, stacking, formation): Mixed. Some patented equipment designs, some process know-how.\r\n- **BMS (Battery Management System)**: Software-heavy, patent + trade secret mix.\r\n- **Recycling processes**: Emerging patent landscape, rapidly evolving.\r\n\r\n### Key Signals\r\n- Battery maker without cathode material patents → likely sourcing from material suppliers\r\n- EV maker without battery patents → definitely sourcing from battery makers (BYD is the exception — vertically integrated)\r\n- Separator / electrolyte additive patents → strong signal of capability depth\r\n\r\n### Verification Strategy\r\n- **Supply agreement announcements**: Battery supply contracts are often publicly announced (e.g., CATL → Tesla/NIO/XPeng)\r\n- **Product certification**: EV models must list battery suppliers for regulatory approval\r\n- **Capacity expansion + customer matching**: When B expands capacity and A launches new models → timing correlation\r\n- **Teardown**: EV battery pack teardowns explicitly identify cell manufacturers\r\n\r\n### Common A-Share Examples\r\n- CATL (宁德时代): Dominant patent portfolio, easily identified as supplier via company announcements\r\n- EV makers (NIO 蔚来, XPeng 小鹏): Zero cell patents → 100% external sourcing\r\n- Putailai (璞泰来), Tianci (天赐材料): Material suppliers → use reverse workflow (their customers are battery makers)\r\n\r\n---\r\n\r\n## 4. Consumer Electronics\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Industrial design / appearance**: Strong design patent coverage\r\n- **Internal components**: Component selection is procurement-driven, not IP-driven. Apple doesn't patent the OLED panel itself.\r\n- **Assembly / integration**: Trade secret (Foxconn's assembly process)\r\n- **Software / UX**: Copyright + trade secret + some patents\r\n\r\n### Key Signals\r\n- **Patent gap is THE NORM in consumer electronics**: A company not patenting components it buys is expected behavior. A patent gap here is weak signal.\r\n- Component teardowns (iFixit, etc.) → the most reliable data source\r\n- Supplier list in annual report → Apple, Xiaomi, Huawei all publish supplier lists\r\n\r\n### Verification Strategy\r\n- **Teardown > supplier list > patent gap**: Reverse priority vs. semiconductor/pharma\r\n- Patent gap in consumer electronics is rarely a meaningful signal — companies consciously don't patent purchased components\r\n- Focus on supplier lists, certification filings (3C certification in China), and teardowns\r\n- **Trade secret assessment**: almost always \"trade secret unlikely\" → but this doesn't mean supply-chain dependency is confirmed; it means patent gap is the wrong signal here\r\n\r\n### Common A-Share Examples\r\n- Xiaomi (小米): Buys panels, chips, cameras → patent gap expected\r\n- Luxshare (立讯精密), Goertek (歌尔股份): Contract manufacturers → their customers are the brand companies (reverse workflow)\r\n- BOE (京东方): Panel maker → use reverse workflow to identify downstream phone/TV brand customers who buy from BOE\r\n\r\n---\r\n\r\n## 5. Materials / Chemicals\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Formulations / recipes**: Classic trade secret domain. Think Coca-Cola, DuPont's Kevlar, 3M's adhesives.\r\n- **Synthesis routes**: Mixed. Some patented (process patents), some trade secrets.\r\n- **Application patents**: How to USE a material → often patented by the user, not the producer. This creates a useful \"reverse\" patent gap signal.\r\n\r\n### Key Signals\r\n- Material user A has application patents for material M but zero production patents for M → A probably buys M\r\n- Material producer B has composition + process patents → strong signal of capability\r\n- High R&D-to-patent ratio → typical for materials companies, not suspicious\r\n\r\n### Verification Strategy\r\n- **Customer audits / qualifications**: Materials for electronics/auto/aerospace require lengthy qualification (1-3 years). Once qualified, switching costs are enormous.\r\n- **Capacity expansion alignment**: When B adds capacity and A enters new markets → timing correlation\r\n- **Trade secret assessment**: Often scores high, requiring procurement/teardown evidence to draw conclusions\r\n- **Financial cross-validation**: Check A's raw material procurement categories against B's product line\r\n\r\n### Common A-Share Examples\r\n- Shin-Etsu (信越化学), SUMCO: Silicon wafer suppliers → their customers are semiconductor fabs (reverse workflow)\r\n- Weihua (万华化学): MDI patents → downstream polyurethane users as customers\r\n- Rongbai (容百科技), Dangsheng (当升科技): Cathode material suppliers → customers are battery makers\r\n\r\n---\r\n\r\n## Industry Selection Quick Reference\r\n\r\n| If the target is... | Priority verification lead | Patent gap signal strength | Trade secret risk |\r\n|---------------------|---------------------------|---------------------------|-------------------|\r\n| Fabless semiconductor | Patent ownership (licensing) | **Medium** | Low |\r\n| Foundry / process | Teardown + financial | **Weak** — process is trade secret | High |\r\n| Innovative pharma | Patent expiry + ANDA | **Strong** — compounds are patented | Low |\r\n| CDMO / CMO | Procurement + financial | **Weak** — processes are trade secrets | High |\r\n| EV maker | Teardown + announcements | **Strong** — battery cells are outsourced | Low |\r\n| Battery maker | Reverse workflow | **Medium** | Medium |\r\n| Consumer electronics OEM | Teardown + supplier list | **Weak** — patent gap is normal | Low |\r\n| Materials producer | Reverse workflow | **Medium** | High |\n\nFile v0.1.1:references/workflow.md\n\n# Detailed Workflow\r\n\r\n## Step 1: Trigger lead identification + direction selection\r\n\r\n**Goal**: Define the analysis target, determine analysis direction, and identify a potential patent gap.\r\n\r\n**Operations**:\r\n1. Confirm target company (full name, ticker)\r\n2. Determine analysis direction based on user intent:\r\n   - **Forward**: user asks \"who supplies X?\" / \"identify suppliers\" / \"what company provides the tech?\" → target is A (the patent-gap company)\r\n   - **Reverse**: user asks \"who are B's customers?\" / \"which downstream companies depend on this?\" / \"what companies need this technology?\" → target is B (the patent owner)\r\n3. Identify the key technology T and/or product P:\r\n   - If not specified by user, extract from product docs, annual report, prospectus, research reports\r\n4. Preliminarily judge patent status:\r\n   - Forward: does A likely lack T's patents?\r\n   - Reverse: does B likely hold dominant/leading T patents?\r\n\r\n**Output**: Direction (forward/reverse) + target company + T + P + preliminary judgment\r\n\r\n## Step 2A: Patent gap confirmation (forward direction)\r\n\r\n**Goal**: Confirm with data whether A truly lacks T-related patents, and rule out legitimate-use scenarios.\r\n\r\n**Operations**:\r\n1. Search A's patents using patent search tools (tushare patent API / PatSnap)\r\n   - Dimensions: A's company name + technology T keywords + IPC class\r\n   - Scope: CN patents + US patents + PCT (per A's market coverage)\r\n   - Note: A may hold patents via subsidiaries / affiliates — search A's related entities\r\n2. Determine gap:\r\n   - A (incl. affiliates) has zero T-related patents → gap confirmed\r\n   - A has T-related patents but count significantly below industry average → partial gap\r\n   - A has sufficient T-related patents → no gap, terminate\r\n3. Rule out legitimate-use scenarios (if gap confirmed):\r\n   - Cross-licensing, acquisition inheritance, patent pool, open-source\r\n\r\n**Output**: Gap determination + excluded legitimate-use scenarios\r\n\r\n## Step 2B: Patent dominance confirmation (reverse direction)\r\n\r\n**Goal**: Confirm that B indeed holds a dominant position in T's patent landscape.\r\n\r\n**Operations**:\r\n1. Search T's patent landscape comprehensively\r\n2. Rank patent holders by quantity and citation count\r\n3. Determine B's dominance:\r\n   - B holds >30% of T's patents OR top-3 cited in T → dominant position confirmed\r\n   - B holds 10-30% → moderate position\r\n   - B holds <10% → weak position, may not be the key technology gatekeeper\r\n4. Identify B's patent family: core patents + timing + jurisdiction coverage\r\n\r\n**Output**: Dominance level (dominant / moderate / weak) + B's core patent list\r\n\r\n## Step 3: Trade secret / know-how assessment\r\n\r\n**Goal**: Determine whether the patent gap reflects true external dependency or deliberate trade secret / know-how protection.\r\n\r\n**Core distinction**:\r\n- **Trade secret**: A has the technology internally, but protects it through secrecy\r\n- **True gap → dependency**: A genuinely lacks the technology and must source it externally\r\n\r\n**Trade secret assessment — 6 indicators**:\r\n\r\n| # | Indicator | Strong trade secret signal | Strong dependency signal |\r\n|---|-----------|---------------------------|--------------------------|\r\n| 1 | Industry norms | Semiconductor process, chemical formula, pharma manufacturing, metallurgy | Consumer electronics, commoditized components |\r\n| 2 | NDA / confidentiality culture | Aggressive NDAs, non-compete lawsuits, \"black box\" secrecy | Open publications, transparent R&D |\r\n| 3 | Patent filing pattern | Files patents in some domains, conspicuously avoids T | Files patents evenly or rarely |\r\n| 4 | R&D intensity vs. patent output | High R&D spend + low T patents | Little to no R&D in T's domain |\r\n| 5 | Technology nature | Process/method/formula (hard to reverse-engineer) | Physical component (identifiable in teardown) |\r\n| 6 | Trade secret litigation | Company/industry trade secret theft lawsuits | Patent litigation is the norm |\r\n\r\n**Scoring**: 4-6 = high likelihood, 2-3 = possible, 0-1 = unlikely.\r\n\r\n**Override rule**: Procurement or teardown evidence can override trade secret finding.\r\n\r\n**Note for reverse workflow**: In reverse direction, spend LESS effort on trade secret assessment for each downstream A — the key question is whether B's patents are essential, not whether A could develop internally. Assessment still runs but reduced weight.\r\n\r\n**Output**: Trade secret likelihood + indicator breakdown\r\n\r\n## Step 4: Multi-lead parallel verification\r\n\r\n**Goal**: Activate 5 leads in parallel. Detailed scoring rules in `checklists.md`.\r\n\r\n### Lead 1 · Patent ownership (confidence: medium)\r\n- Forward: Search T's patent holders, identify primary owner B\r\n- Reverse: For each candidate A, search whether A's products use T and whether A holds T patents\r\n- Check patent status, timeline, NPE risk\r\n\r\n### Lead 2 · Corporate affiliation (confidence: high)\r\n- Search A-B affiliation via Tianyancha / Qichacha / Qixinbao\r\n- Check: mutual investment, JV, executive overlap, historical changes\r\n\r\n### Lead 3 · Talent flow (confidence: medium)\r\n- Search B→A technical-hire / job-change records\r\n- Use WebSearch + job sites\r\n- Multiple technical staff → high likelihood\r\n\r\n### Lead 4 · Procurement (confidence: high)\r\n- Forward: Search A's tender announcements, supplier disclosures for B\r\n- Reverse: Search B's listed customers, annual report client disclosures\r\n- Directly named → high; category only → medium\r\n\r\n### Lead 5 · Product teardown (confidence: very high)\r\n- Physical evidence of B components in A's products\r\n- Teardown reports / BOM analysis / product certifications\r\n\r\n**Output**: Per-lead hit status + per-lead confidence\r\n\r\n## Step 5: Competitive triangulation (NEW v0.3.0)\r\n\r\n**Goal**: Check whether A's industry peers show the same patent gap pattern, providing cross-company corroboration.\r\n\r\n**Why this matters**: A single company's patent gap could be random or internal. But if 3-5 industry peers ALL lack T's patents and ALL point to the same technology owner B, the pattern is unlikely to be coincidental — it strongly suggests B controls essential technology for the entire industry.\r\n\r\n**Operations**:\r\n1. Identify A's direct competitors / industry peers (C₁, C₂...Cₙ):\r\n   - Same industry classification (SW / CITIC sector)\r\n   - Similar product portfolio and revenue composition\r\n   - Target: 3-10 peer companies\r\n2. For each peer Cᵢ, check patent gap for technology T:\r\n   - Search Cᵢ's patent portfolio for T-related patents\r\n   - Record: has patents / has partial patents / has no patents\r\n3. For peers with gaps, check where T likely comes from:\r\n   - Do they also point to the same B? (patent ownership, procurement hints, teardown)\r\n   - Or do different peers point to different owners?\r\n4. Triangulation scoring:\r\n\r\n| Pattern | Interpretation | Triangulation boost |\r\n|---------|---------------|---------------------|\r\n| ≥3 peers have same gap, all point to same B | Industry-wide dependency on B | **+2 levels** (low→high, medium→high) |\r\n| ≥2 peers have same gap, all point to B | Significant corroboration | **+1 level** |\r\n| 1 peer has same gap, points to B | Mild corroboration | No boost, note in report |\r\n| Peers have gaps but point to different owners | Fragmented supply → B's position weaker | **No boost**, flag fragmentation |\r\n| Peers have T patents or different tech paths | A is the outlier → A's gap may be self-inflicted | **No boost**, investigate why A alone lacks |\r\n\r\n5. **Reverse workflow triangulation**: If analyzing B→downstream, check whether multiple unrelated companies in different sub-industries all depend on B's T → the wider the customer diversity, the stronger B's technology-gatekeeper position.\r\n\r\n**Output**: Peer gap matrix + triangulation boost applied + interpretation\r\n\r\n## Step 6: Multi-lead fusion + confidence scoring\r\n\r\n**Goal**: Cross-validate and produce an overall determination.\r\n\r\n**Fusion rules**:\r\n- ≥2 high-confidence leads corroborating → overall \"high\", can conclude\r\n- 1 high + 1 medium corroborating → overall \"medium\"\r\n- Only 1 lead hit, no corroboration → overall \"low\", hypothesis only\r\n- 0 leads hit → terminate, cannot infer\r\n\r\n**Trade secret override rules**:\r\n- Trade secret \"high\" + no procurement/teardown → cap at \"low\"\r\n- Trade secret \"high\" + procurement/teardown confirms → no cap\r\n\r\n**Triangulation boost** (v0.3.0, applied after base scoring):\r\n- +2 levels if ≥3 peers confirm same gap → same B\r\n- +1 level if ≥2 peers confirm same gap → same B\r\n- Triangulation boost CANNOT override trade secret cap\r\n\r\n**Role determination**:\r\n- Leads 4/5 hit (procurement / teardown) → B is A's direct supplier (forward) / A is B's customer (reverse)\r\n- Only leads 1/2 hit → technology licensor or indirect affiliation\r\n- Lead 1 hit but B is NPE → B licenses only, not in supply chain\r\n\r\n**Output**: Overall confidence + triangulation boost applied + B's/A's role\r\n\r\n## Step 7: Financial cross-validation (NEW v0.3.0)\r\n\r\n**Goal**: Cross-validate the inferred supply-chain relationship against hard financial data. No inference should reach \"high\" confidence without financial corroboration.\r\n\r\n**Why this matters**: Patent gaps, corporate affiliations, and even teardowns can point to a relationship that exists but is economically immaterial. Financial data provides the \"size of the bet\" — how much does the relationship matter?\r\n\r\n**Operations** (select applicable checks based on direction):\r\n\r\n### Forward direction (A → find supplier B):\r\n1. **A's supplier concentration** (westock / tushare financial data):\r\n   - Check A's annual report: top-5 supplier list, supplier concentration ratio\r\n   - Is B or a company matching B's profile named?\r\n   - If supplier #1 is an unnamed \"Company X\" matching B's revenue profile → circumstantial\r\n   - Supplier concentration >30% with unnamed top supplier → worth investigating\r\n\r\n2. **A's procurement category matching**:\r\n   - Does A's disclosed procurement categories align with B's products?\r\n   - Check procurement proportion changes over time (growing share = deepening relationship)\r\n\r\n3. **B's customer concentration**:\r\n   - If B is listed: check B's top-5 customer list, customer concentration\r\n   - Does a customer match A's revenue contribution profile?\r\n   - If B has a single large customer contributing >20% of revenue → check if it matches A\r\n\r\n4. **Accounts receivable / payable patterns**:\r\n   - A's AP aging vs B's AR aging — do patterns correlate?\r\n   - Related-party transaction disclosures\r\n\r\n### Reverse direction (B → find customers A):\r\n1. **B's customer concentration**: Who buys from B? Top-5 customer revenue breakdown\r\n2. **B's revenue by industry segment**: Which downstream industries contribute most?\r\n3. **For each candidate A**: Check A's top-5 supplier list for B or B's product category\r\n4. **Related-party disclosures**: Check B and candidates for related-party transactions\r\n\r\n### Financial corroboration scoring:\r\n\r\n| Finding | Corroboration level | Confidence effect |\r\n|---------|---------------------|-------------------|\r\n| B explicitly named as A's supplier (or A as B's customer) | **Strong** | No cap, confirms relationship |\r\n| Procurement category + revenue size matches | **Medium** | Supports but does not confirm |\r\n| Related-party disclosures confirm | **Strong** | Confirms with financial detail |\r\n| No financial evidence found | **None** | **Cap overall confidence at \"medium\"** |\r\n| Financial data contradicts inference | **Negative** | **Reduce confidence by 1 level, flag inconsistency** |\r\n\r\n**Critical rule**: If overall confidence after Step 6 (fusion + triangulation) is \"high\" but financial corroboration is \"none\" → **cap at \"medium\"**. The inference is promising but unverified financially. Only procurement/teardown physical evidence can maintain \"high\" without financial data.\r\n\r\n**Output**: Financial corroboration level + specific evidence + confidence cap applied\r\n\r\n## Step 8: Alternative-hypothesis check\r\n\r\nSee `\n\nArchive v0.1.0: 3 files, 4539 bytes\n\nFiles: skill-card.md (2476b), SKILL.md (6982b), _meta.json (142b)","readmeExcerpt":"Skill: patent-gap-supply-chain Owner: tianzhiceng297-boop Summary: Competitive intelligence workflow that infers supply-chain relationships from patent gaps — with trade secret assessment, competitive triangulation, negative verification, patent quality scoring, evidence freshness tracking, reverse workflow (supplier→client), and financial cross-validation. 通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、 负向验证、专利质量评分、证据时效追踪、反","codeSnippets":[],"executableExamples":[],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\r\nname: patent-gap-supply-chain\r\ndescription: 'Competitive intelligence workflow that infers supply-chain\r\n  relationships from patent gaps — with trade secret assessment, competitive\r\n  triangulation, negative verification, patent quality scoring, evidence\r\n  freshness tracking, reverse workflow (supplier→client), and financial\r\n  cross-validation. 通过\"专利缺口\"反推产业链关系的竞争情报工作流，集商业秘密评估、竞对三角验证、\r\n  负向验证、专利质量评分、证据时效追踪、反向工作流（供应商→客户）、财务数据交叉验证于一体。\r\n  Use when researching tech/hardware/manufacturing/pharma stocks, analyzing\r\n  industry supply chains, identifying suppliers or customers, investigating\r\n  patent gaps, or inferring supply-chain relationships from patent ownership.\r\n  Supports both forward (patent gap → supplier) and reverse (technology owner →\r\n  downstream clients) analysis directions. 触发场景：科技股研究、行业分析、产业链推断、\r\n  供应商识别、客户挖掘、专利缺口分析、专利归属查询、技术依赖分析、上下游推断 / Triggers:\r\n  patent gap analysis, supply chain inference, supplier identification,\r\n  customer identification, technology dependency, competitive intelligence,\r\n  tech stock research, industry analysis, trade secret assessment, negative\r\n  verification, patent quality, financial cross-validation, reverse\r\n  supply-chain analysis. The skill starts from a patent gap (or known technology\r\n  owner), evaluates trade secret likelihood, activates five verification leads\r\n  with competitive triangulation, runs negative verification to rule out false\r\n  positives, scores patent quality, tracks evidence freshness, then\r\n  cross-validates against financial disclosures, outputting a\r\n  confidence-scored supply-chain inference report.'\r\nversion: 0.4.0\r\nagent_created: true\r\ndisable: true\r\n---\r\n\r\n# Patent Gap Supply Chain Inference\r\n\r\n## Overview\r\n\r\nThis skill implements a competitive intelligence workflow: \"patent gap → supply chain inference.\" Core insight: when target company A's product uses technology T, but A holds no patents related to T, there may be a technology-licensing or supply-chain dependency. By searching T's patent ownership to identify company B, then activating multiple verification leads, one can infer whether B has entered A's supply chain.\r\n\r\n**Critical caveat**: A patent gap can also mean A protects T as a **trade secret / know-how** rather than through patents. Patents require public disclosure; many companies (especially in semiconductor processes, chemical formulations, pharma manufacturing) deliberately choose trade secrets. This skill explicitly evaluates this distinction before concluding supply-chain dependency.\r\n\r\n**v0.4.0 new capabilities**:\r\n- **Patent quality scoring**: Not all patents are equal. A core patent (broad claims, high citations, PCT family) carries far more weight than a peripheral patent (narrow claims, single jurisdiction). Lead 1 now scores patent quality, not just quantity.\r\n- **Negative verification**: Actively search for evidence that B is NOT A's supplier (revenue mismatch, capacity mismatch, geography, explicit customer lists excluding A). Only when negativ"},{"path":"README.md","content":"# patent-gap-supply-chain\r\n\r\n通过\"专利缺口\"反推产业链关系的竞争情报工作流 skill，适用于 WorkBuddy / Codebuddy / Claude Code / Codex。\r\n\r\nA competitive intelligence workflow skill that infers supply-chain relationships from patent gaps. Works with WorkBuddy / Codebuddy / Claude Code / Codex.\r\n\r\n---\r\n\r\n## 中文说明\r\n\r\n### 这是什么\r\n\r\n当研究科技股 / 硬件 / 制造 / 医药类公司时，发现标的 A 的产品使用了技术 T，但 A 没有 T 的相关专利——这套 skill 引导从专利缺口切入，**先判断缺口是真实外部依赖还是商业秘密/know-how保护**，再激活五条线索并行验证，竞对三角验证增强信号，**负向验证排除假阳性**，**专利质量评分识别真技术持有者**，**证据时效追踪避免过时结论**，财务数据交叉验证定论，输出带置信度的产业链关系推断报告。\r\n\r\n**v0.4.0 新增**：\r\n- **专利质量评分**：不看数量看质量——权利要求宽度、被引影响、同族规模、法律状态四维评分，核心专利(≥8/12)才支撑高置信度\r\n- **负向验证**：主动寻找\"B 不是 A 供应商\"的证据（收入错配/产品错配/地理不可行/客户名单排除），反证失败正向证据链才成立\r\n- **证据时效追踪**：每条证据带时间戳，>18个月标记为过时并降权，结论不能只靠过时证据\r\n\r\n**v0.3.0**：竞对三角验证 + 财务交叉验证 + 反向工作流 + 行业指南 + 真实案例\r\n**v0.2.0**：商业秘密/know-how 评估\r\n\r\n### 核心设计\r\n\r\n- **商业秘密先排除再定论**：专利缺口可能是商业秘密保护策略，必须先评估\r\n- **专利质量 > 专利数量** (v0.4.0)：一份核心专利胜过几十份外围专利\r\n- **负向验证必做** (v0.4.0)：主动尝试推翻推断，经受住反证的推断才可信\r\n- **证据时效必标** (v0.4.0)：过时证据不能单独支撑高置信度\r\n- **竞对三角验证**：多同行同一缺口 → 同一供应商，是最强信号之一\r\n- **财务交叉验证**：无财报佐证的推断置信度最多\"中等\"\r\n- **双向工作流**：正向（缺口→供应商）+ 反向（专利所有者→下游客户）\r\n- **专利降级为入口线索**：专利缺口只触发验证，不直接定论\r\n\r\n### 安装\r\n\r\n    git clone https://github.com/tianzhiceng297-boop/patent-gap-supply-chain.git ~/.workbuddy/skills/patent-gap-supply-chain\r\n\r\n或手动复制 `SKILL.md` 和 `references/` 到 `~/.workbuddy/skills/patent-gap-supply-chain/`。\r\n\r\n### 触发场景\r\n\r\n- \"研究 XX 科技股\"\r\n- \"分析 XX 公司的供应链 / 供应商\"\r\n- \"查一下 XX 技术的专利归属\"\r\n- \"XX 公司用了某技术但好像没专利\"\r\n- \"识别 XX 的供应商\"\r\n- \"XX 公司的下游客户有哪些？\" (反向工作流)\r\n- \"哪些公司依赖 XX 的技术？\" (反向工作流)\r\n\r\n### 适用边界\r\n\r\n| 标的类型 | 适用性 |\r\n|---------|--------|\r\n| 硬件 / 制造 / 医药 / 材料 | 强适用 |\r\n| 软件 / 互联网 | 弱适用（降低置信度） |\r\n| 商业模式创新 | 不适用 |\r\n\r\n---\r\n\r\n## English\r\n\r\n### What it is\r\n\r\nWhen researching tech / hardware / manufacturing / pharma companies, you find that target A's product uses technology T, but A holds no patents related to T. This skill starts from the patent gap, **first evaluates whether the gap reflects true external dependency or deliberate trade secret / know-how protection**, then activates five parallel verification leads, competitive triangulation, **negative verification**, **patent quality scoring**, **evidence freshness tracking**, and financial cross-validation, outputting a confidence-scored supply-chain inference report.\r\n\r\n**New in v0.4.0**:\r\n- **Patent quality scoring**: 4-factor quality score (claim breadth, citations, family size, legal status). Core patents (≥8/12) support high confidence; peripheral patents cap at low.\r\n- **Negative verification**: Actively search for evidence that B is NOT A's supplier. Inference survives active disproof attempts or is down-weighted.\r\n- **Evidence freshness tracking**: Every evidence item carries a timestamp. Stale (>18 months) evidence is down-weighted; conclusions cannot rest primarily on stale evidence.\r\n\r\n**v0.3.0**: competitive triangulation + financial cross-validation + reverse workflow + industry guide + real case\r\n**v0.2.0**: trade secret / know-how assessment\r\n\r\n### Core design\r\n\r\n- **Trade secrets rul"},{"path":"_meta.json","content":"{\n  \"ownerId\": \"kn718c4g4psbwaq7bt8ft547cd8827z6\",\n  \"slug\": \"patent-gap-supply-chain\",\n  \"version\": \"0.1.2\",\n  \"publishedAt\": 1785984616328\n}"},{"path":"references/checklists.md","content":"# Checklists and Confidence Rules\r\n\r\n## Alternative-Hypothesis Checklist\r\n\r\nFor each inference, check each item and explicitly exclude alternative explanations.\r\n\r\n### Alternative explanations for the patent gap\r\n- [ ] Does A's patent gap stem from a cross-licensing agreement? (check announcements / news)\r\n- [ ] Did A inherit T usage rights via an acquisition? (check M&A history)\r\n- [ ] Does T belong to a patent pool? (e.g., Avanci, MPEG-LA, Via Licensing)\r\n- [ ] Is there an open-source implementation of T? (check open-source projects)\r\n- [ ] Does A hold T patents via a subsidiary / affiliate? (check related-entity patents)\r\n- [ ] Does A deliberately protect T as a trade secret / know-how rather than patenting it? (run Trade Secret Indicators checklist)\r\n- [ ] Is the patent gap simply because T is a purchased component that A has no reason to patent? (common in consumer electronics — weak signal)\r\n\r\n### Alternative explanations for \"B entered the supply chain\"\r\n- [ ] Is B an NPE (non-practicing entity)? If so, B licenses only and is not in the supply chain\r\n- [ ] Does A use B's patent indirectly via a third-party module? The third party is the supplier; B is its upstream\r\n- [ ] Is B only a patent licensor with no physical transaction with A?\r\n- [ ] Is the A-B tie purely financial investment, not business cooperation?\r\n- [ ] Is the A-B relationship economically immaterial? (financial cross-validation required to exclude)\r\n\r\n## Patent Quality Scoring (NEW v0.4.0)\r\n\r\n**Principle**: Patent count is a weak signal. Quality matters. A core patent family outweighs dozens of peripheral patents. Apply this scoring to Lead 1 (patent ownership) and to Step 2B (dominance confirmation).\r\n\r\n### Factor 1 · Claim breadth (0-3)\r\n- [ ] 3: Independent claims covering the core mechanism/architecture, broad scope\r\n- [ ] 2: Claims cover a specific implementation, moderate scope\r\n- [ ] 1: Claims cover a narrow detail/variant\r\n- [ ] 0: Claims are process-only or trivial\r\n\r\n### Factor 2 · Citation impact (0-3)\r\n- [ ] 3: High forward citations (top decile in field); later patents build on it\r\n- [ ] 2: Moderate citations; some later work references it\r\n- [ ] 1: Few citations\r\n- [ ] 0: Never cited\r\n\r\n### Factor 3 · Family size (0-3)\r\n- [ ] 3: PCT + multiple national entries (US/EU/CN/JP/KR)\r\n- [ ] 2: PCT application, few national entries\r\n- [ ] 1: Single jurisdiction (e.g., CN only)\r\n- [ ] 0: Provisional/unpublished\r\n\r\n### Factor 4 · Legal status (0-3)\r\n- [ ] 3: Granted AND maintenance fees paid (active)\r\n- [ ] 2: Granted but recently filed; maintenance status unknown\r\n- [ ] 1: Under examination / pending\r\n- [ ] 0: Lapsed, revoked, abandoned, or transferred to shell\r\n\r\n### Quality levels\r\n- **Core patent**: total ≥8/12 — broad claims, high citations, international family, active legal status\r\n- **Supporting patent**: total 4-7/12\r\n- **Peripheral patent**: total 0-3/12 — narrow, uncited, single-jurisdiction, or lapsed\r\n\r\n### Impact on confidence\r\n- B holds Core patents in T →"},{"path":"references/industry-guide.md","content":"# Industry-Specific Patent & Trade Secret Guide\r\n\r\nDifferent industries have fundamentally different patent/trade-secret dynamics. This guide provides vertical-specific signals, verification strategies, and common pitfalls.\r\n\r\n---\r\n\r\n## 1. Semiconductor\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Manufacturing processes** (lithography, etching, deposition, packaging): Overwhelmingly trade secrets. TSMC, Samsung, Intel do NOT patent their process recipes. The 6-indicator trade secret assessment will almost always score 4+.\r\n- **Chip architectures / designs** (ISA, microarchitecture, interconnects): Strong patent coverage. ARM, x86, RISC-V are heavily patented.\r\n- **EDA tools / design methodologies**: Mixed. Some patented, some trade-secret algorithms.\r\n\r\n### Key Signals\r\n- A foundry's process node advancement without corresponding patents → trade secret, not gap\r\n- Fabless companies (e.g., MediaTek, UNISOC) lacking process patents → expected, they don't manufacture\r\n- Equipment makers (e.g., ASML, AMAT, LAM) → strong patent portfolios, less trade secret reliance\r\n\r\n### Verification Strategy\r\n- **Teardown > everything else**: Chip teardowns (TechInsights, ICWorks) are the gold standard\r\n- **Patent gap in chip design**: meaningful signal for IP licensing (e.g., ARM architecture license)\r\n- **Patent gap in manufacturing**: almost always trade secret, do not infer supplier dependency\r\n- **Financial cross-validation**: check foundry customer concentration in annual reports\r\n\r\n### Common A-Share Examples\r\n- SMIC (中芯国际): Process patents are sparse vs. TSMC; this reflects technology gap + trade secret strategy, not external sourcing\r\n- Hygon (海光信息): x86 license from AMD → genuine IP dependency\r\n- Cambricon (寒武纪): AI chip architecture patents → strong internal IP\r\n\r\n---\r\n\r\n## 2. Pharmaceuticals\r\n\r\n### Patent vs. Trade Secret Dynamics\r\n- **Compound patents** (small molecule structures): Heavily patented. The molecule IS the product.\r\n- **Manufacturing processes** (synthesis routes, purification, formulation): Often trade secrets. Same drug, different manufacturers may use different routes.\r\n- **Clinical trial data**: Regulatory exclusivity + trade secrets, not patents.\r\n- **Biologics / Biosimilars**: Complex mix. Cell line development and culture conditions are trade secrets.\r\n\r\n### Key Signals\r\n- Drug A has expired compound patent but no generic versions → manufacturing process is the barrier (trade secret or new process patent)\r\n- Biologic has zero process patents → expected, cell culture conditions are trade secrets\r\n- ANDA (generic drug application) filings → reveals which companies are trying to enter\r\n\r\n### Verification Strategy\r\n- **Patent expiry + generic entry timing**: The most reliable signal\r\n- **DMF (Drug Master File) filings**: Companies filing DMFs for an API are the actual suppliers\r\n- **CDE (药审中心) registration data**: China-specific regulatory filings\r\n- **Procurement/tender data**: Hospital procurement records, centralized drug procuremen"}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":null,"editorialQuality":{"score":100,"threshold":65,"status":"thin","wordCount":2024,"uniquenessScore":43,"reasons":["uniqueness-below-45"]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-10T21:09:34.170Z","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":"2026-10-10T21:09:34.170Z","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":"2026-10-10T23:50:03.588Z","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"}]}}}