Canoryn

Appearance

Introduction ​

Welcome to the Canoryn documentation.

Canoryn is a local-first agent workspace for macOS. You design workflows on a visual canvas (or by chatting next to it), talk to agents by voice or chat, run live research boards, and optionally let other AI tools author or run the same workflows through the CLI/MCP — on your machine, with your models, under your consent.

What is Canoryn? ​

Canoryn is where Mac workflows become visible and reusable. It is not only a chatbot, and not only a shortcut runner. You build automations as workflows, watch them run, enable them for schedules and external agents, and keep workflows as plain .cryn files on disk.

Core Capabilities ​

📐 Visual Architect + chat ​

Design workflows with a node-based editor — or describe what you want in Chat or Build with chat and watch the workflow appear. Edit one step without rebuilding the whole flow.

🔬 Research boards & live browsers ​

Compare sites in parallel on real Browser Nodes you can scroll and click, synthesize a report, and reopen the board from Finder. See Research boards.

🎙️ Voice, chat & triggers ​

Speak naturally, use chat, or trigger on schedule, app events, webhooks, and more.

🔌 Native app & system actions ​

Control supported apps and system actions from your workflows, with permissions you grant.

🖥️ CLI & other AI tools ​

Install the command-line tool from Settings and connect MCP so Codex, Claude, Cursor, and similar tools can create, enable, and run workflows. See CLI & MCP.

🔑 Bring your own AI ​

Connect your own API keys or run local models. Keys stay in the macOS Keychain.

🔒 Local-first design ​

Workflows, transcripts, and memory stay on your Mac unless you choose a cloud provider for a step.

Next Steps ​