Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Quote & Citation Verifier

    1

    They tidy up wording, merge sentences, or produce a quotation that sounds right but isn't in the document at all.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Quote & Citation Verifier

    Quote & Citation Verifier

    Example session with this skill installed

    You wrote a board summary from our Q3 operations review (source.txt). Before I send it, check the four quotes you used against the original and tell me which ones I can keep.

    • Read your context and instructions
    • Compiled the quote & citation

    I saved the four quotes to claims.json and ran quote-citation-verify against source.txt. Exit code 1: two quotes are fine, one needs a wording fix and one isn't in the source.

    {
      "algorithm": "citelock-v1",
      "nearThreshold": 0.9,
      "sourceSha256": "2797ee12a5c781da781839218e5a1cdb5ae0b2e121e2aa88aa182d877a9e94c1",
      "results": [
        { "quote": "Median first-response time fell from 4.2 hours to 2.9 hours",
    
      "locator": "para 3", "verdict": "VERIFIED", "similarity": 1,
    
          "excerpt": "… new small-business accounts. Median first-response time fell from 4.2 hours to 2.9 hours after the team introduced tri…" },
        { "quote": "We will not expand into new regions until support capacity is stable",
          "locator": "para 5", "verdict": "VERIFIED", "similarity": 1 },
        { "quote": "Customer support volume rose 18 per cent compared with the previous quarter",
    
      "locator": "para 2", "verdict": "NEAR_MATCH", "similarity": 0.9733,
    
          "excerpt": "…y Operations Review, Q3 2026. Customer support volume rose 18 percent compared with the previous quarter, driven mainly by onboarding q…" },
        { "quote": "Churn fell to 2.5 percent, the lowest level on record",
          "locator": "para 4", "verdict": "NOT_FOUND", "similarity": 0, "excerpt": "" }
      ],
      "receiptSha256": "660f679c825593596207ba3d5bcef03f76bf6672f7ec5cd845dede06f38c8dc2"
    }
    

    What to do

    1. Quotes 1 and 2: keep as they are.
    2. Quote 3: the source says "18 percent", not "18 per cent". Use the exact wording if it stays in quotation marks.
    3. Quote 4: not supported. The source says churn "remained flat at 3.1 percent". I've removed the quote and corrected the figure in the draft.
      Receipt: 660f679c…8dc2
    I saved the four quotes to claims.json

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

    About this skill

    Agent Trust: prove the quote is really in the source

    Agents misquote. They tidy up wording, merge sentences, or produce a quotation that sounds right but isn't in the document at all. This skill checks each quote against the source text and gives you a verdict you can show someone.

    What it does

    For each quote you give it, it searches the normalised source text and returns:

    • VERIFIED: the quote appears in the source
    • NEAR_MATCH: 90% or more similar (a typo, a spelling variant, a small paraphrase), with the real excerpt so you can fix the wording
    • NOT_FOUND: no supporting passage found

    When an agent should use it

    • A report, article, legal or research draft quotes a document the agent has
    • You ask "is this quote real?" or want citations checked before publishing
    • A RAG answer quotes retrieved text and you want a deterministic check

    What it outputs

    JSON with the algorithm id, the near-match threshold, a SHA-256 of the normalised source, a per-quote verdict with similarity score and excerpt, and a receipt hash over the whole result. Exit code 0 = every quote VERIFIED; 1 = at least one NEAR_MATCH or NOT_FOUND.

    Limitations

    • It checks that quoted words exist in the source. It does not check that a paraphrased claim is fair or that the source itself is true.
    • The source must be plain text. Convert PDFs and web pages first; bad conversions (broken hyphenation, headers mid-sentence) can lower the match.
    • Similarity is character-based, so a heavily reworded "quote" will come back NOT_FOUND rather than NEAR_MATCH.

    Open source, packaged skill

    Original work by Simon Gant. The core logic is public under the MIT licence (kar-citelock). This listing is the packaged, tested agent skill: SKILL.md workflow, CLI, receipt output and docs. Requires Node.js 18+, zero dependencies, no network, no API keys.

    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 any payment.

    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

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed today

    Requires

    Node.js 18+
    Node.js

    Permissions

    Runs one local Node.js script (`node scripts/quote-citation-verify.js`) through the agent's shell tool. Reads only the source file and claims file you give it. No network access, no file writes, no environment variables read. Pure compute.

    Listedtoday

    What's inside

    Frequently Asked Questions

    No. It checks that quoted wording appears in the source you provide. Whether the source is right is a separate question.

    90% or more similar to a passage in the source, such as a spelling variant or a dropped word. You get the real excerpt so the wording can be fixed.

    Yes: `--quote "text"` checks a single quote without a claims file.

    Not directly. Convert to plain text first, then run it.

    The core logic is MIT open source. You are paying for the packaged, tested agent skill and workflow.

    New in LLM Ops