Watchflows

A local-first, LLM-native visual workflow automation engine for macOS.

Watchflows lets you build visual workflows that react to system events and execute automated actions on your Mac. Combine triggers, conditions, actions, and AI nodes on an infinite canvas to automate anything — no cloud account required.

Node Reference

Triggers

11

Event sources that start workflows: files, apps, schedule, webhooks, keyboard shortcuts, and more.

Emitters

3

Continuous data streams: CPU, memory, network throughput, and per-process metrics.

Conditions

1

Route payloads based on comparisons with Yes/No branching logic.

Transformers

5

Reshape data: loops, templates, math expressions, and key mapping.

Actions

18

Execute operations: scripts, notifications, API calls, file operations, clipboard, and system control.

AI

2

LLM-powered nodes: prompts with structured output and autonomous agents with tool use.

Utilities

7

Debugging and visualization: breakpoints, JSON viewers, live charts, timers, and logging.

Concepts

Payloads

How data flows between nodes using structured payloads and {{variable}} interpolation.

Port Types

The 7 data types, their colors, and coercion rules for connecting ports.

AI Providers

Configure Ollama, LM Studio, OpenAI, Anthropic, Google Gemini, or any OpenAI-compatible endpoint.

Execution Model

How workflows execute: trigger activation, node chains, branching, and completion.

Getting Started

  1. 1
    Download Watchflows from the. Unzip and drag to Applications.
  2. 2
    Create a workflow by dragging a trigger node onto the canvas from the node library.
  3. 3
    Connect nodes by dragging from an outlet port to an inlet port. Data flows through connections as payloads.
  4. 4
    Configure each node in the inspector panel. Use {{variableName}} to reference upstream payload data.
  5. 5
    Run your workflow and watch execution flow through the nodes in real time.
System Requirements: macOS 14.0 (Sonoma) or later.