Signed skill registries
Install verified skill packs. Your host checks signatures and decides which capabilities to trust.
host_requirements
Ed25519 contracts
manifest SHA
trust tiers
Control which tools
AI agents can use.
Nexara is an open-source Rust runtime for AI agent tools. Set permissions, verify skills, and audit every call.
14
Rust Crates
4
Storage Modes
GA
Server APIs
MIT
Licensed
One-line policy
Set tool permissions in one line.
reads orders, never refunds defines what an agent can do. Try the examples below.
Inspectable contract
Catalog simulation
Runtime explanation
Core Capabilities
The contract between AI intent and real-world action.
Policy before execution
Check permissions and require confirmation before a tool runs.
Remote tool contracts
Discover and call remote tools over HTTP with authentication and host allowlists.
Secrets, audit, and learned ranking
Protect secrets and trace tool calls. Optional ranking stays within your policy.
Architecture
Your product stays in control.
You own identity and execution. Nexara handles skill discovery, tool policy, secrets, and audit.
Host
Catalogs, executors, trust resolution, user context, and final action.
Nexara
Registries, runtime policy, secrets, remote contracts, audit, and APIs.
Use Rust or JavaScript.
01 Rust facade
Embed Nexara and enable only the modules you need.
02 Authenticated server
Serve tools and admin APIs behind your authentication.
03 JavaScript SDK
List and call tools from your browser or Node app.
Cargo.toml
[dependencies]
nexara = { version = "0.1", features = [
"runtime",
"registry",
"remote",
"secrets",
"policy",
"server"
] }
JavaScript
import { NexaraClient } from "@nexara/runtime";
const nexara = new NexaraClient({
baseUrl: "https://tools.example.com",
token: process.env.NEXARA_TOKEN
});
const tools = await nexara.listTools({
prompt: "Find the best search capability"
});Start building with Nexara.
Follow the guide to connect your first tools.