rUv is an open-source AI systems builder organized around agent orchestration, adaptive memory, vector infrastructure, and privacy-preserving spatial intelligence. The project maintains 197 public repositories with a verified ecosystem of at least 751 published artifacts across multiple package registries and platforms, including 360 Rust crates, 361 npm packages, 8 PyPI packages, and 22 Hugging Face models and spaces as of July 11, 2026.
The core ecosystem consists of four interconnected flagship systems. RuView provides privacy-first WiFi sensing for presence detection, pose estimation, vital signs monitoring, and environmental intelligence. Ruflo handles multi-agent coordination, swarms, routing, and self-learning workflows at the orchestration layer. MetaHarness generates and evolves portable, verifiable agent harnesses across multiple hosts. RuVector supplies real-time vector, graph, temporal, and adaptive memory infrastructure for learning systems. These are complemented by RVM, which serves as the execution runtime for agentic systems, and emerging research projects including Helix for decision support, RuField for sensor fusion, PhotonLayer for photonic simulation, and WorldGraph for world modeling.
The architecture follows a deliberate layering strategy where RuView and RuField handle perception, Ruflo coordinates multi-agent behavior, MetaHarness generates and evolves agent harnesses, RuVector manages memory and learning, and RVM executes the resulting systems. This connected systems stack approach distinguishes the project from collections of isolated experiments, with each layer building on capabilities provided by lower layers.
The project demonstrates substantial adoption metrics. Cumulative Rust crate downloads reached 1,201,970 as of August 3, 2026, while npm packages accumulated 61,223,077 downloads over the rolling 365-day period from August 3, 2025 to August 2, 2026. Fifteen new owned source repositories were created since June 13, 2026, indicating active ongoing development.
The technology stack spans Rust, TypeScript, WebAssembly, and MCP protocols, enabling cross-platform deployment and interoperability. The project publishes experiments, working systems, package families, and research artifacts publicly to enable inspection, testing, reuse, and dated provenance tracking. The repository includes detailed documentation of project lineage and prior art, with separate dossiers covering account-wide provenance, RuVector internal systems, and the harness-not-model thesis underlying Ruflo and Claude-Flow development.
The project emphasizes transparent provenance documentation, treating repository lineage, first public availability, feature-level evidence, adoption metrics, and novelty claims as separate evidentiary layers. Root commits serve as lineage anchors rather than proof of feature existence at those dates. Comprehensive package indexes, entity glossaries, and machine-readable metrics in the data directory support reproducible claims about the ecosystem's scope and impact. The open publication philosophy prioritizes enabling external verification and reuse of implementations over proprietary control.
