Dafthunk
Dafthunk is an open-source automation platform with a visual editor and a Cloudflare-based runtime. You connect nodes to define a process, then run that process through HTTP requests, scheduled events or other triggers. The project releases its workflow engine, node library and runtime under the MIT license.
Building and inspecting a workflow
A node performs an operation such as an HTTP request, data transformation or AI task. You connect its output to the next node's input to define the data flow. The node documentation describes inputs, outputs and configurable parameters, along with categories for text, images, audio, network operations and documents.
Signed-in users can try individual nodes with sample data in the Playground before adding them to a workflow. This is useful when checking whether a node accepts the data format an upstream step produces. The visual editor also supports live preview and workflow versioning.
Triggers, executions and shared resources
The core concepts distinguish a saved workflow from an execution, which is one run of that workflow. Execution records track status, inputs, outputs and logs, giving you a place to inspect a failed step.
Triggers include HTTP requests, queues, incoming email, cron schedules and Discord or Telegram bot messages. Resources include reusable record schemas, SQLite databases and document datasets for retrieval-augmented generation. Organizations contain these resources and let users collaborate according to their roles and permissions. Third-party services can connect through OAuth integrations, while runtime secrets hold credentials used by nodes.
Deployment and operating costs
Workflows run on Cloudflare Workers and Workflows. Dafthunk uses D1, R2 and Durable Objects for persistent state and describes durable execution that can survive restarts and retries. Its official site lists AI processing, browser automation and media tasks alongside ordinary data and API operations.
You can self-host Dafthunk on your own Cloudflare account. The source license permits modification and commercial embedding, but the account owner remains responsible for the deployment and its costs. The project is provided as-is without warranty. Teams evaluating it should therefore treat the license terms and runtime costs as separate questions and check the nodes and service connections required by their own workflow.