Skip to Content
Naylence Docs are in active development. Share feedback in Discord.
Welcome to Naylence

Welcome to Naylence

Naylence is a federated agentic fabric: a runtime + SDKs for connecting agents and clients across browsers, servers, and containers using logical addressing, routing, and a progressive security model. It’s designed to start simple, and then scale up to production needs like admission control, identity, encryption, durability, flow control, and load balancing without changing your client-facing API.

What you can build

  • Browser + backend polyglot agent workflows (TypeScript ↔ Python)
  • Multi-agent pipelines (fan-out, map/reduce, toolchains, orchestrations)
  • Systems that work across multiple transports (WebSocket, HTTP, broadcast)
  • Deployments that evolve from single process → dedicated sentinel → distributed fabric
  • Secure setups that grow from open/dev → OAuth/OIDC gated → enterprise-grade zero trust

Mental model in 60 seconds

  • A Fabric is the network of connected nodes.
  • A Node is a runtime participant (browser tab, server process, container).
  • A Sentinel is the fabric’s “router”: it anchors connectivity and routes by logical address.
  • An Agent is a callable service (A2A-like API surface).
  • A Client calls agents by logical address; topology changes shouldn’t change client code.
  • Messages move as envelopes (with optional acks, signing, encryption, durability).

Choose your path

If you want to run something in 5 minutes

If you want a complete, runnable walkthrough

If you’re deciding how to deploy

If security is your first concern

Naylence security is progressive—start with the simplest model that meets your needs and add layers as required:

  1. Gated (gated): OAuth2/OIDC admission + token-derived identity
  2. Overlay (overlay): envelope signing (tamper-evidence) using public keys
  3. Advanced (strict-overlay): stronger identity (X.509/SPIFFE) + end-to-end confidentiality

See Security → Overview for the full decision flow and configuration knobs.

When to use Naylence

Use Naylence when you want:

  • a stable logical address model for calling agents across changing deployments
  • multi-transport connectivity (browser + backend realities)
  • a path from “demo” to “production” without rewriting the architecture
  • security that’s configured via profiles and environment variables, not app rewrites

When not to use Naylence

If you only need:

  • a single in-process agent call (no routing, no federation)
  • a single fixed transport with no long-lived network concerns
  • no security, no delivery guarantees, no ops requirements

…a simpler local framework may be enough.

Reference

Last updated on
Overview