Libris · autonomous research · Charon

Libris research demos

Real, fully-cited research reports produced by Charon's autonomous multi-agent research system — each investigates a contested frontier-science question against the live scholarly literature and grades every claim for confidence and evidence strength.

59 sources77 graded claimsevery source cross-checked against CrossRef/arXiv

Reinforcement learning in the human brain

Does the brain use reinforcement learning for motor skill learning and language learning?

30 sources · 38 graded claims · 8 contested

skill learning is textbook dopaminergic RL; the language case is genuinely contested

Gut microbiome & neurodegeneration

Does the gut microbiome causally drive neurodegenerative disease, or merely correlate with it?

8 sources · 8 graded claims · 3 contested

causal evidence is dominated by mouse models while human data is correlational

Epigenetic aging clocks

Do DNA-methylation aging clocks measure a causal aging process, or track correlates?

21 sources · 31 graded claims · judge 8.3/10

central to longevity, sold to consumers, yet causality is unresolved

How each report grades its claims

confidence: high well-supported contradicts evidence against Strong evidence methodological strength verified citation resolved

Confidence (is it true?), stance (does the source support or contradict?), and evidence grade (how strong is the underlying evidence?) are independent axes. Contradicting and contested claims are surfaced, not smoothed over.

Generated by Libris. These are AI-produced research reports — a well-sourced starting point, not a substitute for the primary literature. Citations were verified for existence and title match; not every number was independently fact-checked.