AI Copyright Skill — AI Copyright 技能

AI IP Document Generation

生成 Chinese patent 应用s, software copyright registration materials, or technical disclosure 报告s from AI project code, re搜索 papers, and de签名 docs. Direct Word (.docx) + PPT (.pptx) 输出. Built-in 7 AI domAIn coverage with 11 clAIm templates.

Three 输出 paths:

Patent: CNIPA-格式化 invention patent (clAIms + specification + abstract), with technical disclosure as intermediate deliverable Software Copyright: CPCC registration materials (manual + source code document) Technical Disclosure: From papers/notes to 代理-deliverable disclosure 7 AI DomAIns x 22 Sub-Directions DomAIn Sub-Directions ClAIm Template Perception AI 2D CV · 3D Vision & Graphics · Audio/Video · Sensor Fusion 2.1 Architecture · 2.2 3D Vision Cognitive & Language LLM · Multimodal LLM · RAG · Knowledge Graph 2.3 TrAIning · 2.4 MLLM · 2.5 RAG Generative AI Diffusion & Controlled Gen · AIGC Watermark · Style Transfer 2.6 Diffusion Decision & Interaction 代理 · Embodied AI · RL 2.7 代理 · 2.8 VLA AI Engineering TrAIning/Fine-tuning · Inference Opt · Data Engineering · Edge/IoT 2.9 Inference · 2.10 Data AI Safety & 治理 Adversarial Defense · Watermark · Federated Learning · Alignment 2.11 Watermark Industry 应用s Autonomous Driving · Industry QC · 健康care · Finance · AI4Science · Digital Content DomAIn-adapted Triggers

patent / clAIms / specification / software copyright / disclosure / IP 应用 / paper-to-patent / /AI-copyright / /AI知产

Iteration: When user modifies existing 输出, enter iterative correction flow directly.

Overall Flow Phase 0 Patentability Pre-Assessment (patent path only) Phase A Requirement Diagnosis → path + domAIn + risk level Phase B Project Analysis → auto-检测 (11 types + 6 industries) + 提取 key points Phase C Generation (branch by path) C1 Patent: 搜索 → layout → disclosure → clAIms (11 templates) → specification → abstract → self-检查 C2 Software Copyright: manual (4 templates) → source code doc → self-检查 C3 Technical Disclosure: m应用ing (6 general + 7 domAIn-specific) → drafting → self-检查 Phase D Confirmation Gate Phase E Iterative Correction Phase F Word 输出 (docx-js, auto) Phase G Briefing PPT (python-pptx, patent default)

Phase 0: Patentability Pre-Assessment

Per references/AI-patent-special.md §1, 检查 three elements:

Element 检查 Pass Standard Technical Problem Anchored to specific scenario Not "low efficiency" / "poor accuracy" Technical Means Algorithm steps bound to 系统 architecture Each step linked to HW/SW 组件 Technical Effect Quantifiable Concrete numbers or comparison baselines DomAIn Risk Assessment (§1.3) Risk DomAIn Pitfall Countermeasure HIGH Generative AI (pure content gen) Classified as "rules of mental activities" Must bind to tech scenario + conditional control means HIGH Financial risk control Classified as "business method" Must bind to data processing means HIGH AI alignment / explAInability Classified as "rules of mental activities" Must bind to safety assurance in specific 应用 scenario MED Embodied AI Pure motion control = mental activities Bind each step to sensor 输入 + actuator 输出 MED Re信息rcement Learning Pure strategy optimization = math method Bind to physical 系统 + physics-constrAIned reward MED RAG Pure in格式化ion retrieval = 信息 expression Must show complete technical 流水线 MED AIGC watermark Pure 信息 marking = 信息 expression Watermark embedding must bind to 模型 internal layers

Decision: All pass + low risk → proceed; All pass + high risk → proceed with mandatory domAIn countermeasures; Means fAIl → switch to disclosure path; Effect fAIl → supplement quantitative comparison.

Phase A: Requirement Diagnosis

Confirm with user: path selection, tech topic, AI domAIn (7 domAIns / 22 directions, auto-检测 assisted), 应用licant 信息, inventor 信息, existing materials.

Gate: 输出 3-5 line diagnosis summary (including domAIn attribution and risk level).

Phase B: Project Analysis B.1 Auto-检测ion (§1.1 Decision Tree) Entry file → Project type 应用.py/mAIn.py/serve.py → AI 服务 trAIn.py/trAIner.py → AI TrAIning inference.py/predict.py → AI Inference render.py/gaussian.py/nerf.py → 3D Vision 生成.py/diffusion.py → Generative AI ro机器人.py/vla.py → Embodied AI 流水线.py + langchAIn → 代理 流水线.py + rag/fAIss → RAG 系统 package.json → Frontend/Fullstack

Also 检测 6 industry categories (§1.3): autonomous driving / industry / 健康care / finance / AI4Science / digital content.

B.2 Key Points 提取ion

Priority: 模型 definition → trAIning/inference scripts → rendering/generation/control scripts → papers → de签名 docs → README.

输出: Key Points 列出 (architecture/algorithm/engineering-level innovations, scheme skeleton, key params, distinctions from prior art, quantifiable effects, domAIn attribution).

Gate: Present key points 列出 for user confirmation.

Phase C1: Patent 应用 C1.1 Prior Art 搜索

Online 搜索 2-3 rounds: CN