Remove semantic detours
Agents consume structured node capabilities directly instead of parsing HTML, CSS and interaction layers built for human presentation.
NPS (Neural Protocol Suite) is an open web-infrastructure protocol suite for AI agents and models, unifying communication, node access, identity, discovery and multi-agent orchestration.
The traditional web was designed for browsers and human interfaces. Agents need direct semantics, identity and collaboration boundaries.
Agents consume structured node capabilities directly instead of parsing HTML, CSS and interaction layers built for human presentation.
Reusable schema anchors let high-frequency exchanges reference structure instead of retransmitting the same definitions.
Identity, capabilities and permission scopes are expressed at the protocol layer, giving discovery, invocation and orchestration a shared trust model.
From wire frames to node calls, trusted identity, service discovery and task orchestration, every layer has a focused responsibility and can be implemented independently.
Defines shared frames, encoding, negotiation and transport semantics as the communication foundation of the suite.
View specification ↗Defines discovery, capability manifests and invocation semantics for Memory, Action, Complex, Anchor and Bridge nodes.
View specification ↗Provides verifiable identities, signatures, certificates, capability claims and authorisation scopes for agents and nodes.
View specification ↗Lets nodes announce, resolve and synchronise available capabilities and network topology.
View specification ↗Describes multi-node tasks, dependencies, delegation, retries and result aggregation.
View specification ↗NPS combines language-neutral specifications, machine-readable registries, conformance fixtures and reference implementations across six languages. The LabAcacia repository remains the source of truth for specifications and releases.
LabAcacia is the open-source lab under INNO LOTUS PTY LTD.