Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Citation Quality Inspector

    1

    Audit whether each citation actually supports its attached claim, and flag weak evidence, gaps, and outdated sources.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Citation Quality Inspector

    Citation Quality Inspector

    Example session with this skill installed

    Audit these two claims and references. Claim A: “The intervention reduced missed appointments by 40%.” Citation: Lee et al. (2022), abstract says a pilot improved attendance but gives no percentage in the excerpt supplied. Claim B: “All participants preferred the app.” Citation: same abstract says “most participants found the app easy to use.” We have only the abstract excerpt, not full text. Rate support, explain limits, and rewrite cautiously without inventing details.

    • Read your context and instructions
    • Compiled the citation quality inspector

    Citation support audit

    Scope note: Only the excerpt described in the request is available. The full article, methods, sample, results tables, and citation metadata have not been checked. This is a preliminary source-to-claim review, not a full appraisal.

    ClaimEvidence availableSupport ratingFinding
    A. “The intervention reduced missed appointments by 40%.”Abstract says a pilot improved attendance; no percentage appears in the supplied excerptUnsupported as statedThe direction may be consistent with the excerpt, but the exact 40% figure, denominator, comparator, and meaning of “reduced” are not established
    B. “All participants preferred the app.”Abstract says “most participants found the app easy to use”Not supported; wording overstatesEase of use is not the same as preference. “Most” does not mean all, and the excerpt does not establish preference

    Safer wording

    Claim A: “The abstract reports improved attendance in a pilot study; the supplied excerpt does not provide a percentage effect.”

    • Claim B: “The abstract says most participants found the app easy to use.”

    These rewrites describe only the information reported in the supplied excerpt. They do not establish causality, representativeness, statistical significance, or a population-wide effect.

    Verification needed

    1. Retrieve the full article and confirm the title, authors, journal, year, DOI, and version.
    2. Locate the attendance outcome in the methods and results; record the measure, numerator/denominator, comparator, time period, uncertainty interval, and analysis population.
    3. Check whether the study design supports causal language such as “reduced.” If not, use an associational or descriptive formulation.
    4. Find the exact participant survey item, response count, and wording before making claims about preference.
    5. Note limitations, missing data, and conflicts that materially qualify either result.

    Risk summary

    The numerical claim A is not traceable to the available excerpt and should not be published with “40%.” Claim B changes both quantity (“most” to “all”) and construct (“easy to use” to “preferred”); replace it with the excerpt’s narrower language. Reassess both after the full source is reviewed.

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

    About this skill

    Citation Quality Inspector

    Audit citations in AI answers, research, reports, product comparisons, knowledge assistants, and professional content. Check whether a cited source supports the exact statement—not merely whether a reference exists. Break text into claims, map each claim to source passages, assess relevance, scope, evidence quality, attribution, freshness, and coverage, and identify unsupported or overstated language.

    Distinguish direct support, partial support, context only, contradiction, and no support. Separate what was verified in source text from what remains uncertain. Preserve provenance and cite exact pages or sections when available.

    Never fabricate sources, quotations, page numbers, study results, or verification. A reference listing or abstract alone may not establish what the full source supports. This is a research-quality review aid, not legal, medical, or financial advice.

    Assessment quality depends on source access, context, expertise, and completeness. An abstract may not support a full-text conclusion. The skill cannot independently authenticate documents or replace domain expert review; citation support does not by itself establish truth or overall evidence quality.

    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 1 month ago

    Permissions

    Read Files
    Write Files
    Browser
    Network Access

    Use only source documents the reviewer is authorized to access. Preserve claim-level provenance and source versions. Do not modify or publish source material; require human review for conclusions, especially in regulated domains.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. The exact cited passage must support the claim’s wording and scope. A reference’s existence or general topic match is not enough.

    It can make a limited preliminary assessment, but it must identify that limitation. Full-text conclusions, methods, and numerical results require the relevant source sections.

    No. Support analysis checks how a source relates to a claim. It does not by itself establish truth, study quality, consensus, or applicability.

    New in Research & Analysis