Research workbench
Search live academic sources, read papers in a full-size workspace, build a grounded evidence brief, and choose a contribution that matches your current skill level.
Research query
Cathedral fetches live papers from arXiv, Crossref, PubMed/NCBI, Semantic Scholar, and Europe PMC, then ranks the evidence, exposes the source text, and turns the question into contributable work.
| Paper | Summary | Signal | Source |
|---|---|---|---|
| No papers are loaded into the Cathedral paper layer yet. | |||
Contribution ladder
Cathedral separates useful research work from credentials. Beginners produce traceable evidence; specialists validate interpretation; agents handle repeatable sweeps under human approval.
Read abstracts and verify country, year, K13 mutation, treatment regimen, and measured outcome. No biology training is required.
Deliverable
A checked extraction sheet covering 0 relevant papers from 1 sources.
Validation: A malaria researcher reviews disputed labels; every row retains a source link.
Compare mutation prevalence, delayed-clearance measures, and treatment outcomes across East African locations and years.
Deliverable
A country-by-year evidence table with missing-data flags and contradiction notes.
Validation: Cathedral cross-checks values against the quoted passage and a second reviewer confirms exceptions.
Evaluate whether K13 structure, parasite stress responses, partner-drug resistance, or transmission fitness best explains the observed pattern.
Deliverable
A falsifiable mechanism comparison and the smallest decisive experiment or analysis.
Validation: Domain experts score assumptions, feasibility, and whether the proposed test separates rival explanations.
Delegate citation chasing, mutation normalization, contradictory-result detection, and weekly literature surveillance to Cathedral agents.
Deliverable
A traceable update bundle with sources, extracted claims, uncertainty, and changes since the previous run.
Validation: Nothing becomes project evidence until a person approves the agent output.
Beginner evidence audit
Open each source, record only what it explicitly reports, then mark it checked. Empty cells are useful: they expose what the paper or metadata did not provide.
| Paper | Study setting | Mutation / regimen | Measured outcome | Verified |
|---|
Use this answer
Intelligence layer
Collision alerts
Network-threshold reasoning may translate from traffic flow to horizon information bottlenecks.
Both domains exhibit sharp transitions when local rules accumulate into global regime change.
Contradiction watch
No contradictions flagged yet.
Cathedral will surface cross-paper conflicts once more assumptions are extracted.
Wrong language detector
Percolation and network phase transitions
Clustered nonlinearity suggests threshold behavior rather than independent rider decisions.
Ghost node candidates
Do black holes violate unitarity?
Do black holes violate unitarity?