evidence synthesis

The problem

LLMs frequently hallucinate academic citations, titles, and DOIs, making them unreliable for professional research. Providing a client with a fake reference or a weak study presented as a consensus is a significant professional risk.

What it does

• Authenticates sources against academic indexes like Crossref, OpenAlex, or PubMed to eliminate hallucinations.
• Performs traceable screening by defining explicit inclusion and exclusion criteria for your specific research question.
• Assesses the level of evidence for each source, flagging predatory journals, small sample sizes, or conflicts of interest.
• Synthesizes findings by theme rather than summary, highlighting areas of consensus versus active debate.
• Generates a client-ready executive report with a verified bibliography in APA, IEEE, or Vancouver styles.

Why this beats prompting it yourself

Standard prompts often lead to "summary stacking" where the AI just lists what each paper says without critical weight. This skill encodes a senior researcher's methodology, mandating a verification-first protocol that prevents any unverified data from reaching the final deliverable.

Use cases

• Drafting evidence-based state-of-the-art reports for consulting clients.
• Conducting technical or scientific due diligence on a set of references.
• Automating the screening phase of a literature review with full traceability.
• Verifying the reliability of a bibliography before publication or submission.

Known limitations

Requires web access or academic index integration to perform the mandatory Step 1 verification. It will not simulate verification if the tools are unavailable.