Docs.rs

Edge and vertex blends, chamfers

Assemblies, appearance, PMI, transactions and persistence

Pure, deterministic scheduler core for the Areev agent runtime: graph walk with re-entry generations, journal keys, budgets, and the sans-IO step function. No clock, no randomness, no IO — the driver feeds events, this crate returns commands.

Boolean operations: general fuse and the filters over it

Out-of-box LLM provider backends for Areev and Areev Loop: OpenAI-compatible, Anthropic, and Ollama adapters over a small blocking HTTP client, implementing areev_loop::LlmBackend, plus the free-text-to-Fact extraction used by remember(). Keeps the core crates dependency-light — the HTTP surface lives here.

Shape construction, primitives, measurement, classification and validity

Tessellation: edge discretization and constrained Delaunay triangulation per face

The B-rep data model: shared topology, location chains, orientation, tolerances

Areev Loop: the governed self-improvement engine for AI-agent memory. Standalone engine over an OmsSubstrate (CAL + grains) — zero Areev dependencies.

Core .mg format, canonical serialization, content addressing, and grain types for Areev.

A CLI tool to remove comments from code using tree-sitter for accurate parsing

Curve/curve, curve/surface and surface/surface intersection

Parametric curves and surfaces, 2D and 3D, behind adaptor traits

Geometric primitives, transforms and B-spline basis machinery

Arenas, entity identity, errors, tolerances and geometric predicates for ogeom