tesslops/legal-numbering-architecture

A Claude AI skill that reviews and fixes clause numbering and stale cross-references in legal contracts and agreements.

0.98x

Quality

Pending

Does it follow best practices?

Impact

94%

0.98x

Average score across 10 eval scenarios

Securityby

Pending

The risk profile of this skill

Overview

Eval results

Files

Document Handling

Name: tesslops/legal-numbering-architecture
Rating: 94 (1 reviews)
Author: tesslops

Support for loading and interacting with legal contract documents in two formats: Google Docs (via Chrome integration) and Word documents (.docx via python-docx).

Capabilities

Google Docs

Identifies and accesses Google Docs using Chrome tools. Requires the document to be accessible via URL or already open in Chrome.

Input:
- Google Doc URL, or document already open in Chrome

Process:
- Use Claude in Chrome tools to navigate to the document
- Take a screenshot to orient and confirm correct document
- Scroll through entire document taking screenshots of each page
- Capture all clause numbers via visual inspection

Limitations:
- Requires Chrome integration to be active
- All interactions are visual (screenshot-based)

Usage Example:

When the user provides a Google Docs URL:

Navigate to the URL using Chrome tools
Take an initial screenshot to confirm the document
Scroll through all pages, screenshotting each to build the clause map
Use Edit → Find and Replace (Ctrl+H) for all text changes

Word Documents (.docx)

Accesses .docx files programmatically using the python-docx library.

Input:
- File path to .docx file, or uploaded file

Setup:
- pip install python-docx --break-system-packages

Process:
- Parse full document text programmatically
- Use regex to extract all cross-references
- Access paragraph numId and ilvl properties for numbering analysis
- Apply fixes programmatically across all paragraphs and runs

Usage Example:

from docx import Document

doc = Document("contract.docx")

# Iterate paragraphs
for para in doc.paragraphs:
    print(para.text)
    # Access numbering properties
    if para._element.pPr is not None:
        numPr = para._element.pPr.numPr
        if numPr is not None:
            ilvl = numPr.ilvl.val
            numId = numPr.numId.val

# Find and replace text
import re
pattern = re.compile(r'clause \d+(\.\d+)*', re.IGNORECASE)
for para in doc.paragraphs:
    for run in para.runs:
        matches = pattern.findall(run.text)
        # process matches

doc.save("contract_fixed.docx")

Document Type Identification

If user hasn't specified document type:
- Ask for document: Google Doc URL or .docx file path/upload
- Google Doc: treat as Chrome-based visual workflow
- .docx: treat as programmatic python-docx workflow

tesslops/legal-numbering-architecture