一键在 Manus 中运行任何 Skill

patent-prior-art

星标3

分支0

更新时间2026年7月2日 06:00

Real patent prior-art search for the examiner panel. Generates patent-search queries from an invention topic/claims, queries a keyless patent search (Google Patents query endpoint; ddgs site:patents.google.com fallback), and LLM-filters genuine on-topic prior art. Produces a prior-art-of-record reference list that patent_reviewer consumes via --prior-art, so §102/§103 rejections are grounded in named real references instead of fabricated-from-memory citations. Complements literature_surfacer (which covers academic non-patent literature).

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

BoltzmannEntropy

BoltzmannEntropy/QuantumNovelty

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

patent_prior_art

Grounds the USPTO examiner panel in REAL patent prior art. Adapted from the ScienceSkills prior-art search pattern (code_prior_art query-gen + relevance filter; novelty_check keyless web search).

Outputs: prior_art.json (full records), prior_art_refs.json (flat list of relevant publication numbers — feed to patent_reviewer --prior-art), and prior_art.md.

Usage: run.sh --topic "..." [--claims FILE] [--max 20] [--llm claude] [--no-llm] --output DIR

同仓库更多 Skills

同仓库

qn-chain

BoltzmannEntropy/QuantumNovelty

Modular workflow chain composing the QuantumNovelty skill catalog (deep_research, quantum_reviewer, logical_fallacies, novelty_audit, cross_llm_prediction, pareto_explorer, quantum_paper, process_summary, ...) into named pipelines for quantum-computing research. Per-stage `--with-X` / `--skip-X` toggles, structured telemetry (`pipeline_summary.json`, `decision_history.json`, `_stage_health.json`, `HEARTBEAT_AUDIT.md`, `checkpoint.json` — AutoResearchClaw's stage-health pattern), strict no-fallback env defaults, and a cross-model falsifiability fork. The QN-specific contributions are: paper-audit as a first-class pipeline, the 11 quantum-CS-specific fallacy taxonomy in the fallacies stage, the cross-LLM amplitude-prediction skill (`cross_llm_prediction`) as an opt-in stage, and the audit-and-falsify Pareto novelty check.

2026-07-043

patent-reviewer

BoltzmannEntropy/QuantumNovelty

Simulated USPTO examiner panel for quantum-computing patents. Ingests a Google Patents URL / publication number / saved file, examines every claim individually under 35 U.S.C. §§ 101/102/103/112 + a quantum-operability check, runs claim-compliance and full-application review under USPTO/EPO/PCT standards, and emits an Office Action plus a deterministic machine-actionable `_office_action.json` (disposition, rejected/allowed claims, rejections-by-statute). The patent analogue of quantum_reviewer.

2026-07-043

patent-drafter

BoltzmannEntropy/QuantumNovelty

Guided quantum-patent drafting workflow. Starts from an invention disclosure or short topic and drafts a filing package for USPTO, EPO, or multi-standard review: title, abstract, background, summary, drawings plan, detailed description, examples, claims, claim-compliance notes, application formalities, and prosecution checklist.

2026-06-273

name

patent_prior_art

description

patent_prior_art

Grounds the USPTO examiner panel in REAL patent prior art. Adapted from the ScienceSkills prior-art search pattern (code_prior_art query-gen + relevance filter; novelty_check keyless web search).

Outputs: prior_art.json (full records), prior_art_refs.json (flat list of relevant publication numbers — feed to patent_reviewer --prior-art), and prior_art.md.

Usage: run.sh --topic "..." [--claims FILE] [--max 20] [--llm claude] [--no-llm] --output DIR