plagiarism-engine

Detect verbatim and paraphrased plagiarism using HNSW indexing and fingerprinting.

1|Updated Feb 28, 2026

Search Tags:#vector search #document analysis #similarity #originality #plagiarism #winnowing

Ready to run

npx skills add https://github.com/Patrickjoshanedez/CMS-V2 --skill plagiarism-engine

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: plagiarism-engine
Source: https://github.com/Patrickjoshanedez/CMS-V2/tree/main/.agents/skills/plagiarism-engine
Command: npx skills add https://github.com/Patrickjoshanedez/CMS-V2 --skill plagiarism-engine

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires chromadb, sentence-transformers, bullmq, ioredis, @xenova/transformers, @qdrant/js-client-rest, celery, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill provides robust solutions for detecting plagiarism and ensuring the originality of submitted documents, safeguarding academic integrity and intellectual property.

Core Features & Use Cases

Verbatim Plagiarism Detection: Identifies direct copy-pasted content using Winnowing and Rabin-Karp algorithms.
Paraphrased Plagiarism Detection: Finds semantically similar content using HNSW vector indexing and sentence embeddings.
Use Case: An academic institution can use this Skill to automatically scan student submissions against a vast archive of existing documents, flagging any instances of plagiarism with detailed reports and highlighted matches.

Quick Start

Use the plagiarism-engine skill to check the originality of the provided text against the project archive.

Frequently Asked Questions about plagiarism-engine

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I detect paraphrased plagiarism in academic documents?▼

Detect paraphrased plagiarism by using HNSW vector indexing with sentence embeddings to find semantically similar content. This approach identifies rewritten text by comparing meaning rather than exact string matching against existing document archives.

What is the best way to identify verbatim copy-pasted content in student submissions?▼

The best way to identify verbatim copied content is using Winnowing and Rabin-Karp fingerprinting algorithms. These methods create document fingerprints to efficiently match exact text strings across large archives of existing submissions.

Can I use ChromaDB or Qdrant for document similarity analysis at scale?▼

Yes, you can use ChromaDB or Qdrant for document similarity analysis. They provide HNSW vector indexing to handle high-performance semantic comparisons across large-scale document archives asynchronously via BullMQ or Celery task queues.

How do I check document originality against an existing project archive?▼

Check document originality by submitting text to scan against your project archive. The engine uses vector search and fingerprinting to flag both verbatim and paraphrased matches, generating detailed reports with highlighted similarities.

Does plagiarism detection work with sentence-transformers for semantic matching?▼

Yes, plagiarism detection works with sentence-transformers to generate sentence embeddings for semantic matching. These embeddings power the HNSW vector indexing required to accurately identify paraphrased content within submitted documents.

What are the limitations of vector search for plagiarism detection?▼

Vector search limitations for plagiarism detection include missing exact verbatim matches if used alone. It must be combined with Winnowing fingerprinting to ensure both direct copy-pasted and semantically paraphrased text are accurately identified.