Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Grandma's Wisdom

    1

    A rigorous academic audit tool to verify AI-generated claims, detect hallucinated citations, and check source fidelity.

    $9.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Grandma's Wisdom

    Grandma's Wisdom

    Example session with this skill installed

    Audit this claim and citation: "Recent studies (Smith et al., 2023) prove that daily caffeine intake increases long-term memory by 45% in adults." Reference: Smith, J. (2023). Caffeine and Brain. JNS.

    • Read your context and instructions
    • Compiled the grandma s wisdom

    The audit of the provided claim and citation reveals significant fidelity issues and a potential citation hallucination.

    Academic Claim & Citation Audit

    1. Audit Summary

    • Claims reviewed: 1
    • Citations reviewed: 1
    • Verified items: 0
    • Unsupported claims: 1
    • Potentially fabricated citations: 1
    • Items requiring human review: 1

    2. Claim-Level Audit

    IDClaimVerification TargetEvidenceStatusConfidenceAction
    C-01Caffeine increases long-term memory by 45%Effect size / Source existenceNo matching record for Smith (2023) in JNSUnverifiableHighVerify source existence

    3. Citation Identity Check

    CitationExists?AuthorsTitleVenueYearStatus
    Smith (2023)NoSmith, J.Caffeine and BrainJNS (Journal of Neuroscience?)2023Hallucination Risk

    4. Source-Fidelity Check

    ClaimCited SourceSupporting EvidenceFidelityIssue
    45% memory increaseSmith et al., 2023None foundZeroSource not found; effect size highly improbable for nutritional studies

    6. Bullshit Detection Flags

    • Fabricated citation: "Smith (2023) JNS" does not match any indexed paper for this year/volume.
    • False precision: The "45%" figure is a common hallmark of LLM hallucination in academic contexts.
    • Overgeneralization: Claims "proof" without qualifying variables.

    8. Correction Recommendations

    Original claim: Recent studies (Smith et al., 2023) prove that daily caffeine intake increases long-term memory by 45% in adults.

    • Problem: The citation is likely hallucinated; the effect size is physiologically implausible.

    Safer revision: Research suggests a correlation between caffeine and short-term cognitive tasks, though long-term memory effects remain under debate.

    • Citation change: Replace with a verified meta-analysis (e.g., Nehlig, 2010 or similar).

    Next steps

    • Verify if "JNS" refers to Journal of Neuroscience and search their 2023 archives manually.
    • Search for the specific 45% statistic in PubMed to identify the actual source if it exists.
    • Remove the Smith (2023) reference from the bibliography until a DOI is provided.

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

    About this skill

    AI models often hallucinate scholarly citations or overstate the strength of academic evidence. This skill performs a rigorous, claim-level audit of AI-generated academic content to ensure every statement is backed by a verifiable source. It moves beyond simple plagiarism checks by verifying the actual relationship between a claim and its cited evidence.

    What it does

    • Claim extraction identifies atomic, checkable statements within a text block.
    • Citation verification confirms the bibliographic existence of cited works using identifiers like DOI or PMID.
    • Source-fidelity analysis checks if a cited source actually supports the specific claim attached to it.
    • Overstatement detection flags when causal language, effect sizes, or populations exceed the source evidence.
    • Fabrication spotting isolates hallucinated papers, invented quotations, and citation stacking.

    How it works

    1. Input analysis involves parsing the AI-generated text, bibliography, and any provided source documents.
    2. Bibliographic resolution resolves each citation against scholarly records to verify existence and metadata accuracy.
    3. Evidence mapping compares each extracted claim against the supporting passages in the cited literature.
    4. Status assignment categorizes every claim and citation using a structured audit table.

    Frameworks & tools

    This skill is designed for scholarly workflows involving DOI, PMID, and arXiv identifiers. It operates on academic manuscripts, literature reviews, and white papers.

    Why this beats prompting it yourself

    Standard prompts often fail to distinguish between a citation that exists and a citation that actually supports a claim. This skill enforces a multi-step verification pipeline that separates bibliographic validity from claim support, preventing "plausible-sounding" academic hallucinations.

    Use cases

    • Auditing AI-generated literature reviews for hallucinated citations.
    • Verifying the accuracy of evidence in technical white papers.
    • Pre-submission checks for academic manuscripts to ensure citation fidelity.
    • Fact-checking AI summaries of complex medical or scientific research papers.

    Known limitations

    Requires access to citation metadata or source documents to verify existence. Cannot verify claims where the primary source is behind an inaccessible paywall or not provided in the context.

    Requires access to citation metadata or source text to verify support. Cannot confirm claims from paywalled or unavailable primary sources.

    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

    No special permissions declared or detected

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    Unlike standard plagiarism checkers that only look for copied text, this skill analyzes the logical relationship between a claim and its cited source to ensure the evidence actually supports the statement made.

    The skill is compatible with major LLM orchestration platforms and can be integrated into workflows handling scholarly manuscripts, literature reviews, and technical white papers.

    While the skill can resolve identifiers like DOI and PMID to verify a paper's existence, it requires either provided source text or access to open-access databases to confirm the specific claims within a paper.

    You receive the full skill logic, integration documentation, and access to future version updates that improve citation resolution and hallucination detection patterns.

    Setup is straightforward: simply add the skill to your agent's workflow and ensure your input text includes standard academic identifiers (like DOI or arXiv IDs) for the most accurate verification results.

    New in Research & Analysis