← Back to Skills Marketplace
Biaoshu Writer
by
wuliwenjing
· GitHub ↗
· v5.4.0
· MIT-0
332
Downloads
2
Stars
1
Active Installs
7
Versions
Install in OpenClaw
/install biaoshu-writer
Description
标书撰写器 v5.4.0 - 投标技术标文档自动生成工具。支持解析 txt/pdf/docx/xlsx 招标文件，生成符合评分标准的技术标 Word 文档。适用：技术标编写、交通工程（高速/航道）投标。
README (SKILL.md)
标书撰写器 v5.4.0
投标技术标文档自动生成工具。发送招标文件 → 自动生成符合评分标准的技术标 Word 文档。
更新日志
v5.4.0 (2026-04-14)
- 🐛 修复：
merge_chapters.py确保章节间空行分隔，修复章节不连续问题 - 🐛 修复： 禁止章节末尾写「本章小结」，避免一级标题导致结构错乱
- 🐛 修复： 章节序号统一为中文数字，禁止混用阿拉伯数字
- ⚙️ 优化： 新增
merge_chapters.py合并脚本，替代直接 cat 拼接
v5.3.0 (2026-04-01)
- 🐛 修复： Word 文档元数据创建时间从 2013 年修正为当前时间；creator/description 改为 User
- ⚙️ 优化：
convert_to_word.py保存后自动修复 docx 元数据
v5.2.1 (2026-03-27)
- 新增
check_chapter_words.py章节字数检查脚本
环境安装
依赖库
pip install python-docx pdfplumber openpyxl PyPDF2
字体
将 SimSun.ttf（宋体）复制到 ~/Library/Fonts/（从 Windows C:\Windows\Fonts\ 复制或网上下载）
执行流程（标准）
① 发送招标文件 → ② 解析评分标准与采购需求
③ 生成4级标题详细大纲 → Owen 审核大纲
④ 审核通过后派发子进程并发编写各章节
⑤ 章节字数检查（必须步骤）
⑥ humanizer-zh 去 AI 痕迹（必须步骤）
⑦ 汇总整合 → 转换 Word
字数检查（必须）
公式：目标字数 = 评分分值 × (总页数 ÷ 总分) × 780
合格范围：目标 × 0.75 ~ 1.25
| 分值 | 目标字数 | 合格范围 |
|---|---|---|
| 5分 | 20,000字 | 15,000~25,000字 |
| 4分 | 16,000字 | 12,000~20,000字 |
不达标 → 打回对应章节重写 → 达标后继续
格式参数
格式指令（Markdown 顶部加注释块）
\x3C!-- doc-format
font: SimSun
body-size: 16pt
title-level: 36pt
sub-level: 32pt
line-spacing: 26pt
margins: 2cm
first-line-indent: 0.74cm
-->
默认值
| 参数 | 默认值 | 说明 |
|---|---|---|
| font | SimSun | 正文字体 |
| body-size | 16pt | 正文字号（三号） |
| title-level | 36pt | 一级标题 |
| sub-level | 32pt | 二级标题 |
| line-spacing | 26pt | 行距 |
| margins | 2cm | 页边距 |
| first-line-indent | 0.74cm | 首行缩进（2字符） |
常用模板
| 模板 | 参数 |
|---|---|
| 政府标书 | SimSun / 16pt / 26磅 / 2cm |
| 高速公路 | SimSun / 16pt / 28磅 / 2.5cm |
| 航道工程 | SimSun / 16pt / 26磅 / 2cm |
内容编写规则
| 规则 | 说明 |
|---|---|
| 段落结构 | 每小节（### X.X.X）≥ 3 个独立段落，禁止单一长段落 |
| 表格展示 | 每个章节尽量包含表格，不能全是文字 |
| 禁用词汇 | ❌ 不用"我方/我们"；✅ 用"将/项目组/本方案" |
| 金额描述 | ❌ 禁止在技术标中出现金额/预算描述 |
| 内容贴合 | 严格按评分标准和采购需求编写，不泛泛而谈 |
| 章节结尾 | ❌ 禁止写「本章小结」「本章小结」「本章总结」等章节总结段；章节正文写完即结束，禁止添加任何以 # 开头的总结性标题 |
| 章节序号 | ✅ 统一使用中文数字（一、二、三…），禁止混用阿拉伯数字（第3章 vs 第三章） |
核心脚本
| 脚本 | 功能 |
|---|---|
parse_bid_files.py |
解析 txt / pdf / docx / xlsx 招标文件 |
convert_to_word.py |
Markdown → Word（含元数据修复 v5.3.0） |
check_chapter_words.py |
章节字数检查 |
check-font.sh |
检查 SimSun.ttf 字体是否安装 |
install-deps.sh |
一键安装 Python 依赖 |
输出规范
- 输出路径：
/Users/owen/Desktop/{项目名称}/ - Word 文件命名：
{项目名称}_技术标_{日期}.docx - 项目目录模板：
/Users/owen/Desktop/{项目名称}/
└── 章节/
├── 01_第一章.md
└── {项目名称}_技术标_{日期}.docx
Usage Guidance
This package appears to do what it says (parse tender files, enforce chapter word counts, merge chapters, and convert Markdown to Word). Before installing or running it: 1) Review the scripts locally and run them inside an isolated environment (create a VM or container) and use the provided install-deps.sh which creates a virtualenv. 2) Fix known code issues first (e.g., check_chapter_words.py has a malformed print statement that will crash the script). 3) Ensure your chapter filenames match the naming convention expected by merge_chapters.py (numeric prefix like '01_...') or adapt the script. 4) Be cautious when following instructions to copy/download SimSun.ttf (font licensing and source trust); prefer obtaining fonts from legitimate sources. 5) The 'humanizer-zh' step intentionally removes AI-writing traces — consider legal/ethical/compliance implications before using it in regulated or audit-sensitive bids. 6) Do not run the tools on sensitive or confidential documents until you've validated behavior on non-sensitive samples. If you want, I can point to the exact lines that need fixes and suggest safe test commands to run in a temporary directory.
Capability Analysis
Type: OpenClaw Skill
Name: biaoshu-writer
Version: 5.4.0
The biaoshu-writer bundle is a specialized tool designed for automating the generation of technical bid documents. It includes scripts for parsing various file formats (PDF, Word, Excel), calculating word counts based on scoring criteria (check_chapter_words.py), and converting Markdown to Word with professional formatting (convert_to_word.py). The bundle features a 'humanizer' prompt (humanizer-zh.md) to improve AI output quality and a metadata-cleaning function in the Word converter to ensure document anonymity, which are standard requirements in professional bidding. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found.
Capability Assessment
Purpose & Capability
Name/description match the provided scripts: parse_bid_files.py (txt/pdf/docx/xlsx parsing), convert_to_word.py (Markdown→docx), check_chapter_words.py (word-count checks), merge_chapters.py, and font checks. Declared pip dependencies (python-docx, pdfplumber, openpyxl, PyPDF2) align with those functions.
Instruction Scope
SKILL.md stays within the stated purpose (send tender files → parse → generate chapters → humanize → check words → merge → convert). It references user-local paths (~/Library/Fonts/SimSun.ttf and /Users/owen/Desktop/{项目名称}/) which is expected for document output, but be aware these are hardcoded defaults the scripts/prompts assume. The included 'humanizer-zh' module is explicitly for removing AI-writing traces — functionally aligned but potentially ethically sensitive. There are code-quality issues: check_chapter_words.py includes an obvious printing bug (a malformed print: 'print(f=" * 60")') that will throw an error and break the 'must' word-check step if run unmodified; some parsing/sorting assumptions (e.g., filenames parsed by numeric prefix) can fail on unexpected filenames.
Install Mechanism
Install spec is a standard pip install of well-known PyPI packages. install-deps.sh creates a virtualenv and installs those packages. No downloads from arbitrary URLs, no archive extraction, and no obscure third-party installers detected in the provided files.
Credentials
The skill declares no required environment variables, no credentials, and no config-path access beyond typical user font and Desktop paths. That is proportional for a local file-parsing / document-generation tool.
Persistence & Privilege
The skill is not marked always:true, does not request persistent elevated privileges, and does not modify other skills' configurations. It operates on local files and a virtualenv; autonomous invocation is allowed by default but is not a special privilege here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install biaoshu-writer - After installation, invoke the skill by name or use
/biaoshu-writer - Provide required inputs per the skill's parameter spec and get structured output
Version History
v5.4.0
修复章节合并/小结/序号三个问题
v5.3.1
精简优化 SKILL.md：去除冗余版本历史表；补充输出规范目录模板；流程更清晰；内容规则更结构化
v5.3.0
修复Word文档元数据创建时间2013年问题；creator/description改为User；convert_to_word.py保存后自动修复docx元数据
v5.2.1
v5.2.1: SKILL.md重写优化，body-size统一16pt
v5.2.0
v5.2: 表格表头白色填充，清理标题括号，水平线跳过
v5.0.1
v5.0.1: 添加格式确认步骤，生成前询问用户字体/字号/行距偏好
v5.0.0
v5.0: Support custom formatting, PDF OCR detection, venv isolation, INSTALL.html guide
Metadata
Frequently Asked Questions
What is Biaoshu Writer?
标书撰写器 v5.4.0 - 投标技术标文档自动生成工具。支持解析 txt/pdf/docx/xlsx 招标文件，生成符合评分标准的技术标 Word 文档。适用：技术标编写、交通工程（高速/航道）投标。 It is an AI Agent Skill for Claude Code / OpenClaw, with 332 downloads so far.
How do I install Biaoshu Writer?
Run "/install biaoshu-writer" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Biaoshu Writer free?
Yes, Biaoshu Writer is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Biaoshu Writer support?
Biaoshu Writer is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Biaoshu Writer?
It is built and maintained by wuliwenjing (@wuliwenjing); the current version is v5.4.0.
More Skills