Documentation
Noren is an open, agent-native operating layer. It runs locally on the host and exposes eight operating-layer primitives (intent, runtime, memory, identity, permissions, comms, compositor, and settlement) through which human users, software agents, and tools coordinate work, share state, and settle usage.
The documentation is organized into concepts, architecture, reference, protocols, and operations. The Getting started guide covers installation and an end-to-end intent dispatch; Concepts establishes the model referenced throughout the remainder of the documentation.
Browse by area
Status
Noren is pre-1.0 systems infrastructure. The local daemon, CLI, TUI, identity, permissions, memory, audit, peer auth, budget, and local receipt ledger are implemented in the repository. MCP, A2A, the local web console, autonomous workflow records, live coverage matrix, and provenance envelopes are actively hardening. Runtime isolation includes manifest-level requirements, trusted-local fail-closed behavior, daemon-selectable Linux gVisor configuration, runtime-level gVisor runner work, live Linux sandbox coverage, and a documented Linux runner setup. Noren's settlement chain is Robinhood Chain (chain id 4663, USDG, ETH for gas). Solidity contracts (escrow, bond-receipt and reputation verifiers, trade guards) exist in the repository with Foundry tests, but nothing is deployed on-chain yet; deployment is coming soon. The Noren token contract is deployed (see the token page). A separate x402 gateway and facilitator for USDG is already live at gateway.usenoren.xyz. Distributed settlement, cross-chain enforcement, installers, additional-language SDK publication (Python and Rust-crate), release-scope and audit-root signing, and transparency publication are tracked on the roadmap.
Source repository: coming soon.
Position in the stack
Noren operates between the host operating system and user-facing agentic applications. It provides identity, permissions, memory, communication, and settlement as shared, host-level services, allowing language models, agent frameworks, and end-user applications to integrate against a common substrate rather than reimplement these primitives independently.