Clear answers for every step of your review.

What Coventra includes, what it does not do, and where human review responsibility starts and ends.

What Coventra records, exposes, and exports so teams can review how an analysis was produced without relying on a black-box result.

Create a project, choose sensible settings, invite the right people, and prepare your workspace before records arrive.

Detailed guidance on workspace entry, project creation, metadata, analysis configuration, outcomes, treatments, DOI lookup, file imports, and PDFs.

Use search/import tools without flooding the project, preserve search history, and keep imported records ready for screening.

Run two-phase screening, understand blind mode, use reviewer assignments, and know what happens when decisions disagree.

Use phase-aware conflict queues, threaded decision comments, and durable resolution without overwriting original reviewer judgments.

Everything in the Screening workspace: imports, deduplication, search logs, settings, decisions, assignments, conflicts, reliability, active learning, and PRISMA reporting.

Upload PDFs, use the evidence viewer, fill the extraction grid, and preserve source-backed provenance for manuscript tables and analysis.

A detailed guide to PDFs, evidence viewing, extraction rows, provenance, harmonization, blind extraction, MCP suggestions, data freeze, and gated AI suggestions.

Choose the right analysis family, inspect model settings, understand premium gates, and prepare interpretable outputs.

Detailed guidance for pairwise analysis, saved runs, diagnostics, advanced analyses, NMA, Bayesian methods, GRADE, generated text, and interpretation limits.

Use generic quality templates, assessment rows, comments, conflicts, and plots without relying on bundled proprietary instruments.

A guide to quality templates, assessment records, comments, conflicts, validation, plots, and how to bring in external assessment instruments safely.

Invite the right access level, assign project tasks, understand categorized work queues, and keep team communication auditable.

Every collaboration concept explained: full-project access, screening-only members, premium passes, categorized tasks, notifications, comments, and what each user should expect.

Know which exports to use for manuscripts, screening reports, extraction data, analysis reproducibility, and long-term audit trails.

Every available export explained, including what it contains, when to use it, and which formats require premium project access.

Practical guidance on accounts, project access, PDFs, AI suggestions, exports, and responsible handling of research data.

How Coventra MCP connects an agent to a review project with project-scoped consent, proposal-only writes, PDF-grounded evidence, and human checkpoints.

Results from Coventra's PDF-grounded MCP extraction benchmark: protocol-guided outcome targeting, source-backed proposals, repeat-run protection, and human review checkpoints.

Methods and workflow results from the first Coventra MCP benchmark run: title/abstract screening, full-text progression, and PDF-grounded extraction proposals in a real oncology SRMA project.

A transparent explanation of why meta-analysis software can run successfully while still producing statistically wrong results, and how review teams should check methods before publication.

How Coventra records numerical validation evidence by comparing app outputs with direct R package results, what the current validation library covers, and what remains a human methodological responsibility.

A transparent map of Coventra's default statistical assumptions, which analysis choices users can change, and which assumptions must be verified before journal use.

Formula-level documentation for pairwise synthesis, including binary, continuous, generic, and time-to-event data, sparse-data handling, heterogeneity, defaults, and changeable model settings.

How Coventra renders study effects, pooled estimates, weights, confidence intervals, prediction intervals, and journal-style display options.

Formula-level and interpretation guidance for leave-one-out checks, influence diagnostics, cumulative meta-analysis, funnel plots, Egger, Begg, Peters-style tests, contour guides, and trim-and-fill sensitivity output.

How subgroup fields and study-level covariates connect to moderator analyses, including formulas, defaults, limitations, and reporting requirements.

Coventra's Bayesian pairwise workflow, including exact bayesmeta integration, default priors, changeable prior settings, prior sensitivity, and diagnostic limits.

Detailed methodology for contrast-based and binary-arm network meta-analysis, including consistency equations, defaults, changeable settings, and transitivity checks.

How network plots represent treatments, direct comparisons, multi-arm evidence, and connectivity without implying effect magnitude or certainty.

How Coventra produces and interprets treatment rankings, including ranking direction, uncertainty, P-scores, and rankogram limitations.

Methodology for presenting network estimates versus a reference treatment and all pairwise network comparisons in numeric and heatmap form.

How comparison-adjusted funnel plots are generated for network meta-analysis and why sparse or heterogeneous networks limit interpretation.

Network inconsistency and confidence-support outputs, including direct-vs-indirect comparison, loop requirements, contribution matrices, and certainty caveats.

Detailed methodology for pooled single-arm proportions, incidence rates, and raw means, including formulas, GLMM defaults, person-time units, heterogeneity, funnel limitations, and reporting requirements.

How reported estimates and standard errors are entered, transformed, pooled, and checked when arm-level data are insufficient or inappropriate.

Design-specific formulas and extraction requirements for studies that should not be treated as ordinary independent-arm trials.