Boo AI Technical Bid Number Continuity Checker

Problem

Technical bid documents in Word often develop skipped, repeated, or misaligned numbers as sections are inserted, split, or reordered. Manual review is usually limited to visual scanning: it is hard to confirm which outline level a paragraph belongs to, and harder to distinguish a legitimate restart from a true numbering break. This skill treats the .docx file as OOXML structure and turns numbering continuity into inspectable, annotatable issues.

How It Works

The workflow produces three artifacts:

• Outline extraction: extract_outline() reads outline levels from both direct paragraph settings and style inheritance. OutlineLvl values run from 0 for the top level to 8 for nine levels, producing _大纲目录.txt.
• Numbering analysis: analyze_numbering() groups paragraphs by outline level, then checks continuity within the same level. It recognizes common patterns such as 一、, 1、, (1), and 1., and uses subsequence splitting to separate restarts from breaks. Table numbering is counted separately and attached to the enclosing outline node.
• Highlighted comments: generate_highlighted() marks broken paragraphs with paragraph-level yellow shading, creates word/comments.xml for comment definitions, and updates [Content_Types].xml plus relationship files.

It relies on Python 3.x standard libraries such as zipfile, xml.etree.ElementTree, re, and os, and does not require pandoc or python-docx.

Limits

This skill is best used to inspect the numbering structure of an existing .docx technical bid. It outputs an outline directory, a numbering analysis, and an annotated document. It is not a general proofreader, does not rewrite prose, normalize terminology, or replace templates. Highly custom numbering, fields in headers or footers, and non-native Word hierarchy may still require manual confirmation.