# Deterministic skills: Skills that give the same answer every time

These skills and modules compute their results with ordinary code, so your agent gets the same answer for the same input and can check it. Every one here carries its own tests and declares no network access.

A model can explain a Brier score; a tested function computes it. Handing arithmetic, date math and parsing to code keeps model calls for the steps that need judgment, and the result can be checked against the component's own known answers.

## Groups

- [Statistics for comparing results](https://baltor.ai/deterministic-skills/statistics): Calibration, conformal intervals, anytime-valid confidence sequences, shrinkage and kernel two-sample tests as tested Python modules with known answers.
- [Checks for agent work](https://baltor.ai/deterministic-skills/agent-work-checks): Typed work outcomes, independent evidence counts, equal-cohort comparisons and append-only ledgers that check agent work with code instead of a second model.
- [Calculators for public statistics](https://baltor.ai/deterministic-skills/public-statistics): Rates, ratios and indices from published definitions, with confidence bounds and explicit denominators: mortality, poverty, food security, vaccination and more.
- [Records, schedules and budgets](https://baltor.ai/deterministic-skills/records-and-schedules): Critical paths, earned value, rosters, appointment slots and retention dates as pure functions of JSON input, each with a known good and a known wrong example.
- [Algorithms that show their work](https://baltor.ai/deterministic-skills/algorithms): Principal components without NumPy, sorting and union-find traces, RDFS and OWL 2 RL closures, collision clearance and render plans, with schemas and tests.
- [Geometry and mesh algorithms](https://baltor.ai/deterministic-skills/geometry): Pure Python triangulation, convex hulls, polyline simplification, offsets, subdivision, curves and gears whose glTF output loads in Godot.

Every component is the exact version the Baltor library release of October 9, 2026 serves. Your agent downloads it through Baltor with Pro, $49 a month: https://baltor.ai/get-started