👾 Version 2026.3.12 · Node.js 22.12+ · Playwright-powered

Memory-nativeagent runtime.

Two-layer memory, retrospective learning, Playwright browser control, and a gateway control plane in one runtime. Stateful across CLI and messaging surfaces.

Open Quick Start Read Getting Started

Two-layer memory Retrospective learning Playwright runtime Gateway control plane

👾

Agent runtime Two-layer memory, browser control, and gateway services

Layer A stores machine-readable profile data. Layer B writes a desktop mirror at ~/Desktop/resonix-M/. Browser and gateway services sit beside it.

Core surfaces

memoryLayer A profile + Layer B desktop mirror

browserplaywright-core with resilient automation flows

gatewayrouting, auth, status, and channel delivery

Memory mirror

~/Desktop/resonix-M/
identity/
knowledge/
autonomy/
retrospectives/
logs/

Operational visibility

Success rate, P95 latency, and risk signals stay visible.

Health probes Gateway status Automation board

Two-layer permanent memory

Layer A stores profile data. Layer B writes readable memory files to the desktop mirror.

Retrospective learning loop

Each task feeds review, learning, and the next execution path.

Gateway observability

Track success, errors, P95 latency, and risk level in one place.

Why it feels different

Built for continuity.

Continuity

Memory is part of the architecture, not an afterthought.

Identity, knowledge, retrospectives, and logs persist so the next interaction starts with retained state instead of a blank session.

Learning

Retrospectives feed the next execution path.

What worked, what failed, and what should change all feed back into the runtime.

Operations

Browser execution and channel delivery share one runtime contract.

Browser control, gateway routing, and channel adapters live in one runtime contract.

Resonix vs traditional assistants

Persistent, not session-bound.

Traditional assistants

Memory Context window only

Learning Starts fresh each thread

Identity Mostly generic

Browser Usually external tools

Resonix

Memory Permanent + desktop mirror

Learning Retrospective feedback loop

Identity Profile-driven state

Browser Built-in Playwright runtime

System architecture

Channel ingress, gateway control, persistent state.

Channels

Telegram, Discord, Slack, WhatsApp, Signal, plus extension channels.

Gateway

Routing, auth, status tracking, and channel adapters form the control plane.

Agent runtime

Memory, browser execution, tools, identity, and growth loops run together.

Quick Start

Install. Onboard. Serve.

curl -fsSL https://raw.githubusercontent.com/mangiapanejohn-dev/Resonix-AG/main/install.sh | bash

iwr -useb https://raw.githubusercontent.com/mangiapanejohn-dev/Resonix-AG/main/install.ps1 | iex

curl -fsSL https://raw.githubusercontent.com/mangiapanejohn-dev/Resonix-AG/main/install-termux.sh | bash

After install

resonix -v
resonix onboard
resonix gateway

resonix gateway status
resonix cron board
resonix memory profile

Use onboard for setup, gateway for serving, gateway status for health, cron board for schedules, and memory profile for learned state.

Documentation

Docs for operators.

Docs for install, channels, memory, providers, and security.

Start here for setup, routing, memory, and production-facing operational detail.

Open docs hub

Getting Started Installation verification, onboarding, and health checks. Install One-line bootstrap commands plus source-install references. Channels Gateway-backed messaging surfaces and rollout strategy. Configuration Environment configuration and deployment-aware defaults. Memory Two-layer memory architecture and long-term recall. Security Risk controls and production-facing safeguards.