How does this skill prevent AI hallucinations in academic citations?

The skill generates a structured JSON evidence ledger and a Markdown report that maps your claims directly to verbatim excerpts, preventing the 'citation drift' common in standard AI summaries.

Can I use this skill if my source documents are in a different language than my final report?

It performs cross-language auditing by analyzing source documents in their original language and providing the final audit report in your preferred language without losing technical nuance.

What specific files and tools are included with the 'lb evidence audit' purchase?

You receive the core AI agent logic, a specialized JSON schema for evidence validation, and a set of Python 3.11 scripts to verify the structural integrity of your audit reports locally.

Are there any specific technical requirements for running the local verification scripts?

While the agent logic handles the analysis, the bundled Python scripts require a local environment with Python 3.11+ and the jsonschema library to perform the final verification step.

Does this skill verify the scientific accuracy of the sources provided?

This skill focuses on structural and literal consistency, meaning it verifies that a claim is backed by a specific excerpt, but it does not evaluate the underlying scientific validity or truth of the source material itself.

Academic Citation & Evidence Auditor

Example session with this skill installed

Audit the following three PDF sources regarding carbon sequestration in peatlands and generate an evidence ledger in English. Identify any claims that lack direct verbatim support.

Read your context and instructions
Compiled the academic citation

I have audited the provided documents on carbon sequestration in peatlands. I traced each assertion to a verbatim excerpt and mapped them to the required schema.

Evidence Audit Results

Claim ID	Academic Assertion	Source Reference	Evidence Status
CLM-001	Peatlands store 30% of global soil carbon.	Source A, p. 12	Verified (Literal: "peatlands... hold ~30% of all soil carbon")
CLM-002	Drainage-induced oxidation is irreversible.	Source B, p. 5	Gap (Source indicates "long-term damage," not irreversibility)
CLM-003	Sphagnum growth rates peak at 15°C.	Source C, p. 89	Verified (Literal: "optimum thermal window near 15°C")

{
  "audit_metadata": {
    "version": "2.0.0",
    "language_mode": "multilingual-on-request",
    "status": "instructor_review_required"
  },
  "ledger": [
    {
      "id": "CLM-001",
      "source_excerpt": "Although covering only 3% of the land surface, peatlands... hold ~30% of all soil carbon.",
      "citation_id": "Ref-2023-A",
      "integrity_check": "pass"
    }
  ]
}

Next steps

Review the evidence_report.md for the full context of the identified evidentiary gaps.
Run python scripts/verify_evidence.py evidence.json to confirm schema compliance.
Provide additional sources for CLM-002 to resolve the "irreversibility" claim status.

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

About this skill

See what a citation actually supports

Academic Citation & Evidence Auditor connects academic claims to exact excerpts from the sources you provide. It helps make unsupported assertions, overextended conclusions and missing evidence visible before a lecture or technical draft is distributed.

Readable findings, inspectable records

The workflow prepares an audit report and structured evidence records. Preserve the original excerpt, source identifier and claim relationship so a reviewer can inspect the underlying material instead of trusting a general statement that a citation was checked.

With multilingual sources, original quotations remain in their source language. Translated explanations are separate, so a polished translation does not replace the evidence being reviewed.

Understand the levels of checking

The local scripts validate structural references and literal text consistency. Those checks can establish that a referenced excerpt exists in the supplied text; they cannot establish that the source is scientifically correct or that the excerpt logically supports every interpretation.

The agent can assist with assessing support and identifying gaps, but those judgments remain review items. The product is an evidence-auditing workflow, not a scientific truth certificate, plagiarism detector or substitute for expert review.

Use it before publishing or teaching

Supply a lecture draft, the relevant source texts and your desired report language. Review the claim-to-source map and address unsupported material before proceeding. Choose the full Lecture Builder Suite when you also need planning, drafting, assessments, visuals and Word publishing.

Text-based review requires a compatible agent and readable sources. Python 3.11+ and jsonschema are needed for the local validation helpers. Source access, PDF extraction and web research depend on the host and are not included services.

How to use

Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

Reviews

No reviews yet

Be one of the first to try it. Every listed skill passes our trust checks below.

Security scanned

Passed 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.

Academic Citation & Evidence Auditor