A focused sample size calculator for biostatisticians designing clinical trials. Public-release scope: implements the standard textbook methods every biostat student and working scientist needs, in a clean parameterization-and-execution pattern that drops into a Claude Code workflow. Built by Vera and battle-tested in pharma R&D before being released as an open-source skill.

What it does

Endpoint	Single-arm	Controlled (1:1)
Binary	Exact binomial	Z-test unpooled
Continuous	One-sample t-test	Two-sample t-test
Time-to-event	Exponential rate	Schoenfeld log-rank

Outputs a sample size table (CSV) across a configurable alpha × power grid and a power-vs-N curve (PDF). Base R only — no external packages required.

What it does not do

These are intentionally out of scope and listed in the skill’s Beyond This Skill section with primary references:

Bayesian Go/No-Go decision frameworks
Pre-trial assurance (PPOS) for Phase 2 → Phase 3 transitions
2:1 (or other unequal) randomization
Poisson sample size for incidence-rate endpoints
Unconditional exact (Barnard-type) tests
Frequentist post-trial decision analysis
Operating-characteristics simulation across true-parameter grids

If your design needs any of those, treat this skill’s output as a baseline.

Who it’s for

Biostatisticians producing a first-pass sample size table for a study protocol or design discussion.
Clinical scientists validating a CRO’s calculation or running a quick sensitivity check.
PhD / MS holders entering biostat roles who want a clean reference implementation that matches the conventions taught in standard texts.

Tested

Yes — built by a working biostatistician for internal use, then trimmed to a public-scope subset before release. The full internal version is preserved on the v1.0-internal git tag.

Verdict

The skill structures execution. Endpoint selection, H0/H1 parameterization, and regulatory defense remain the biostatistician’s responsibility — and that’s the point.