Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Similarity Preflight

    1

    Rapidly triage batches of student submissions to flag identical text overlap and shared paragraph structures.

    Free

    1 useSecurity scanned
    Similarity Preflight

    Similarity Preflight

    Example session with this skill installed

    /similarity-preflight on the ./submissions folder. Check if students copied from each other.

    • Read your context and instructions
    • Compiled the similarity preflight
    • Generated the data export

    I have scanned 12 files in the ./submissions folder. I normalized the text and performed a cross-check for identical blocks and shared sequences exceeding 12 words.

    Similarity preflight — ./submissions · 12 files

    Verdict: 2 pairs to review

    PairShared evidenceWhereCall
    smith_essay.pdf ↔ jones_essay.pdf"The underlying mechanism of photosynthesis in C4 plants involves..."p.2 / p.1REVIEW
    smith_essay.pdf ↔ jones_essay.pdfidentical introductory paragraph (55 words)p.1 / p.1REVIEW
    miller_lab.docx ↔ davis_lab.docx"results were consistent with the initial hypothesis"p.4 / p.4CLEAR
    wilson_final.txt ↔ key_ref.md8 shared words in bibliography—CLEAR

    CLEAR: 10 files with no notable overlap or only common phrasing.

    Next steps

    1. Review the shared paragraphs in smith_essay.pdf and jones_essay.pdf to determine if the overlap is coincidental.
    2. Run assignment-integrity-screener if you require a full CSV report with similarity percentages and ranked scores.
    3. Check if the shared phrasing in miller_lab.docx is due to a standard lab template.

    Read-only. No scores, no grades, no verdict on intent.

    smith_essay.pdf

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Manual plagiarism checks are slow and prone to oversight when scanning large batches of student submissions. This skill provides a read-only triage to identify exact and near-exact text overlap before a formal integrity review.

    What it does

    • Batch scanning extracts text from folders containing .txt, .md, .pdf, and .docx files.
    • Text normalization strips whitespace and punctuation to detect obfuscated copying.
    • Overlap detection identifies identical paragraphs and long shared sequences of 12 or more words.
    • Structural analysis flags identical sentence skeletons where only specific names or nouns have been swapped.
    • Evidence mapping records the specific shared snippets and their locations within each file for human verification.

    How it works

    1. Inventory: The agent lists all submissions and identifies if a reference answer key is present.
    2. Normalize: It processes the text into a searchable format while maintaining a map to the original page numbers.
    3. Compare: It executes a cross-check between all files or against the provided key to find matching phrases.
    4. Report: It generates a summary table of findings with clear REVIEW or CLEAR calls for each pair.

    Frameworks & tools

    This tool relies on internal text extraction for PDF and Word documents and standard string normalization logic. It is environment-agnostic and requires no external software installations.

    Why this beats prompting it yourself

    Writing a prompt to compare dozens of files usually hits context window limits or results in hallucinations regarding which file contains which text. This skill uses a structured normalization and mapping workflow to ensure evidence is quoted accurately from specific page numbers.

    Use cases

    • Identify two students who may have shared a draft before submission.
    • Check a batch of essays against a provided reference answer key.
    • Spot copy-pasted paragraphs that have been moved to different sections.
    • Perform a quick sanity check on 20+ files before starting the grading process.

    Known limitations

    It cannot provide similarity scores, percentage rankings, or automated grading. The skill is read-only and will skip files it cannot extract text from without attempting a repair.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Education & Learning.

    Read-only triage only. Cannot provide similarity scores, percentage rankings, or plagiarism verdicts. Does not support OCR for scanned images within PDFs.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    1 use

    Times used to date

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 26 days ago

    • Free with an Agensi account

    Permissions

    Read Files
    Write Files

    File Scopes

    **/*.txt
    **/*.md
    **/*.pdf
    **/*.docx
    Listed26 days ago

    What's inside

    Frequently Asked Questions

    The skill identifies direct text matches, near-exact overlaps, and structural copying (where sentence skeletons are identical but nouns are changed), but it does not scan external internet databases or websites.

    Unlike simple prompts, this skill uses a structured mapping workflow that bypasses context window limits, preventing the AI from 'hallucinating' or mixing up text between different students' files.

    The skill supports .txt, .md, .pdf, and .docx formats; however, it cannot read scanned images or handwritten notes, as it relies on text extraction rather than OCR.

    No, this is a 'read-only' triage tool designed to identify evidence for human review; it does not assign grades, calculate percentage scores, or issue formal disciplinary reports.

    This is a standalone, environment-agnostic skill that integrates with any AI agent capable of file system access, requiring no external software installations or API keys.

    New in Education & Learning

    Free