How does semantic comparison differ from a standard text diff?

Standard diffs highlight character changes, whereas semantic comparison identifies changes in meaning, requirements, and logical flow, even if the text is completely reworded.

Can this skill be used for legal or highly technical documents?

Yes, it is specifically designed to handle complex relationships like 'IF-THEN-ELSE' and 'Step A must follow Step B' found in technical specs and procedural manuals.

Why does this skill run extraction tasks in parallel?

Parallel execution prevents 'priming' the AI with knowledge from the first document, ensures independent discovery of facts, and reduces total processing time by nearly 50%.

What is an execution equivalence score?

It is a weighted metric (40% claims, 40% relationships, 20% graph topology) that determines how similar the functional outcome of following the two documents would be.

How reproducible are the comparison results?

By using temperature=0 and pinned model versions, the skill achieves high reproducibility with a typical variance of only ±1-2% due to inherent LLM semantic judgment.

Semantic Document Comparison

Name: Semantic Document Comparison
Author: cowwoc

bycowwoc

•

内容管理

Compares two documents semantically to identify functional and structural differences while preserving logical relationships.

关于

The compare-docs skill allows Claude to perform deep semantic audits between two documents, moving far beyond simple text-based diffs. By independently extracting atomic claims and complex logical relationships—such as temporal dependencies and conditional branches—in parallel, it determines if the underlying meaning and execution flow remain identical even if the wording or organization has changed. This is essential for developers, technical writers, and analysts who need to verify that refactored documentation, updated specifications, or rewritten procedures maintain their original logical integrity and requirement coverage.

主要功能

Comprehensive relationship mapping covering temporal, conditional, hierarchical, and exclusion constraints.
Independent discovery workflow that prevents confirmation bias by extracting data from both sources before comparison.
Parallel extraction of claims and relationships to ensure zero cross-contamination and 50% faster processing.
Execution equivalence scoring based on claims (40%), relationships (40%), and dependency graphs (20%).
Deterministic analysis settings using temperature 0 and model pinning for reproducible results.
2 GitHub stars

使用场景

Auditing procedural guides to ensure step-by-step logic and dependencies remain intact after content updates.
Verifying that a refactored technical specification or API documentation preserves all original requirements.
Comparing versioned policy documents or legal contracts to identify subtle changes in conditional clauses.

关于

主要功能

Comprehensive relationship mapping covering temporal, conditional, hierarchical, and exclusion constraints.
Independent discovery workflow that prevents confirmation bias by extracting data from both sources before comparison.
Parallel extraction of claims and relationships to ensure zero cross-contamination and 50% faster processing.
Execution equivalence scoring based on claims (40%), relationships (40%), and dependency graphs (20%).
Deterministic analysis settings using temperature 0 and model pinning for reproducible results.
2 GitHub stars

使用场景

Auditing procedural guides to ensure step-by-step logic and dependencies remain intact after content updates.
Verifying that a refactored technical specification or API documentation preserves all original requirements.
Comparing versioned policy documents or legal contracts to identify subtle changes in conditional clauses.