agentleFS
KumnXiKumnXi/patent-aid/.claude/skills/idea-to-disclosure/SKILL.md

idea-to-disclosure

Skill2 starsChanged 3 months agoMITView on GitHub

Convert a technical idea into a submission-ready Chinese patent disclosure (docx). Three-stage LLM generation with anti-hallucination checks and claim-format validation. Works for any technical domain; built-in examples include power-grid and pipeline-inspection robotics. use when 用户提供技术想法/问题/方案并要求\"生成交底书\"\"写技术交底\"或一键出 Word,从想法起步而非从交底书起步

The file

Files
SKILL.md

想法 → 技术交底书

项目核心目标:从想法到技术交底书。当用户提供技术想法(而非现成交底书)时,按此流程执行。

输入字段

字段说明必需
idea技术想法描述✅
title发明名称("一种...")建议
tech-field技术领域可选
purpose要解决的问题可选
core-method核心方法/技术路线可选
problems现有技术不足可选

想法若含"针对某缺陷…用某方法…达到某效果"结构即可直接生成;缺字段可让用户补充或用默认。

执行步骤

  1. 确认想法已收集:从用户描述中提炼 idea(含缺陷/方法/效果要素)
  2. 若用户贴图(照片/手绘/系统框图/论文截图/PDF):先经 VL"看图"提炼字段,再进入流水线。图片转成结构化字段后才交给 DeepSeek(src/api/vl_client.py 文字桥接):
# 单图或多图:输出 {idea, title, tech_field, purpose, core_method, problems}
D:/Anaconda3/envs/mathmodel/python.exe -m src.utils.idea_vision 图1.png 图2.png
  • 图片内容(设备结构/流程/数值)会被提炼成 idea 字段,缺的字段留空让用户补或走默认
  • 未配置 DashScope / 读图失败时自动回退纯文本输入,不阻断流程
  1. 调用流水线(内部自动执行三阶段 LLM 生成 → 防无中生有修复 → 权利要求校验 → 合规审查 → 加密保存历史 → Word 导出):
D:/Anaconda3/envs/mathmodel/python.exe scripts/generate_patent.py "技术想法" \
    --title "一种..." --tech-field "技术领域" \
    --purpose "要解决的问题" --core-method "核心方法" \
    --problems "现有技术不足" --out output/交底书.docx

3b. 交付自查(可选,交付前建议跑):Word 导出后渲染回看显示质量(公式/附图/布局/乱码),由 VL 看图质检,避免"导出成功但渲染有问题":

# 生成后直接自查;或对已有 docx 单独自查
D:/Anaconda3/envs/mathmodel/python.exe scripts/generate_patent.py ... --selfcheck
D:/Anaconda3/envs/mathmodel/python.exe scripts/selfcheck_disclosure.py output/交底书.docx
  • 原理:docx → LibreOffice 转 PDF → 逐页 PNG → qwen-vl 回看(见 src/utils/disclosure_inspector.py)
  • 需 LibreOffice(winget install --id TheDocumentFoundation.LibreOffice);未装时提示安装,不阻断生成
  • 终端显示每页问题数;有 error 级问题建议修复后重新导出

输出检查(生成后核对)

  • 终端应显示:生成模式(llm_staged 为最优)、质检评分(>90 为好)、防无中生有修复数、权利要求校验摘要
  • Word 内含附图:mermaid 流程图经 Graphviz dot 渲染(300dpi),见 src/utils/diagram_generator.py
  • 生成历史已加密保存到 data/disclosure_history/(见 src/utils/history.py)
  • --selfcheck 后终端应显示每页自查结论(P01 正常 / 检出 N 个问题)

注意事项

  • 首次运行引擎初始化约 1 分钟(建知识图谱/RAG 索引)
  • 全文约 1.2-1.9 万字,三阶段模式质检平均 96 分
  • 需要真实 DeepSeek API Key(config/api_config.json,保密,不入库)
  • 若用户手上有现成交底书而非想法 → 引导用 /patent-writer 进入撰写阶段

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.