Connect Claude, ChatGPT, OpenAI Codex, and Cursor directly to the Analysez AI biostatistics engine. Every distribution check, Table 1 summary, and p-value is computed by verified statistical code — never guessed by a language model.
CLAUDE / CODEX / GEMINI / CURSOR (MCP)
https://app.analysez.com/api/mcpCHATGPT CUSTOM GPT ACTIONS (OPENAPI 3.1)
https://app.analysez.com/api/mcp/openapi.jsonlist_studiesreadOnlyHint: trueBrowse Your Studies
Lists the authenticated researcher’s studies on Analysez AI, including study ID, title, workflow step, and whether a cleaned dataset is ready.
scan_dataset_healthreadOnlyHint: trueCheck Your Data Quality
Runs a thorough, read-only data quality check on a study or inline CSV dataset. Spots missing values, impossible clinical or survey entries, coded missing values, mixed Arabic/English categories, duplicates, and direct identifiers without modifying your data.
clean_studyreadOnlyHint: trueInspect Study Cleaning Status and Decisions
Reports where your study is in the cleaning workflow and shows what needs your decision next—with real row evidence, recommended fixes, deep review proposals, and missing-data options. Never modifies data.
apply_cleaningreadOnlyHint: trueApply Approved Study Cleaning Decisions
Applies only the cleaning fixes, deep review cell edits, or missing-data handling choices that the researcher explicitly approved. Always saves a new cleaned dataset version and audit log while keeping your original uploaded file untouched.
generate_table_1readOnlyHint: trueBuild Publication Table 1
Computes baseline descriptive statistics (Table 1) with distribution checks, Mean ± SD, Median (IQR), and categorical counts and percentages (category rows are marked `is_level: true`), optionally grouped by study arm.
select_defensible_testreadOnlyHint: truePick the Right Statistical Test
Checks variable types, sample sizes, data distribution, and group structure to recommend a defensible statistical test before running analysis.
run_verified_statisticsreadOnlyHint: trueRun Statistical Tests and Models
Runs a verified hypothesis test or regression model (two-sample t-test, paired t-test, Mann-Whitney U, Chi-Square, correlation, or linear regression) and returns exact test statistics, p-values, 95% confidence intervals, and effect sizes.
audit_manuscript_numbersreadOnlyHint: trueCheck Draft Manuscript Numbers
Cross-checks sample sizes (N), averages, percentages, and p-values written in a draft Results paragraph against the actual numbers computed from your study dataset.
generate_methodology_defensereadOnlyHint: truePrepare Methodology Justification and Reporting Checklist
Produces a clear justification for the selected statistical test, compares alternative parametric and non-parametric tests, computes minimum detectable effect size at 80% power, and generates a CONSORT or STROBE reporting checklist.
verify_summary_statisticsreadOnlyHint: trueVerify Summary Statistics and Table Consistency
Recalculates t-statistics, degrees of freedom, Chi-Square, Odds Ratios, p-values, and mean granularity consistency (GRIM) from published summary numbers without requiring raw participant rows.