BrainChain
BrainChain is the visual, node-based pipeline that orchestrates the work behind a Mushin project — taking your design, running it through the AI, and turning it into structured metadata. You build and run it in the BrainChain editor, and you can watch every step as it happens. This page explains how it works.
Nodes and connections
A BrainChain pipeline is a graph of nodes, each doing one job, connected by edges that pass data and control between them. Nodes cover the steps of the generation process — loading design streams, running them through the AI, splitting and joining work across features, and producing the structured output.
Connections attach to typed, color-coded handles on each node, so you can see at a glance what kind of thing flows where — the flow of control, the design streams, the differences, and the feature information each move between nodes along their own handles. A run always begins at a start node and follows the connections from there.
For most work you do not build a pipeline from scratch — your project comes with one — but the editor lets you inspect it and, when you need to, adjust it. Reusable sub-pipelines can be collapsed into a single module node and dropped into other pipelines.
Running the pipeline
BrainChain runs as jobs. There are three kinds, matching the stages of turning a design into a specification:
| Job | What it does |
|---|---|
| Plan | Works out the order the pipeline’s steps must run in |
| Execute | Runs the pipeline, calling the AI and producing the design output |
| Build Meta | Turns the produced design into structured metadata — the knowledge base your app is generated from |
You start a job from the editor. Because a run coordinates real work, one pipeline runs at a time for a project.
Watching a run
While a job runs, the Log Viewer streams its progress live, so you can follow what each node is doing and see the results flow through. You can open the detail of any step to see exactly what went in and came out. When a run finishes, the full log stays available to review, page by page.
This visibility is a large part of BrainChain’s value: generation is not a black box. You can see the pipeline plan, watch it execute, and inspect every node’s contribution — which makes it far easier to understand what the AI produced and to trust the result.
Related pages
- The Neuro-Symbolic Approach — the design the pipeline works from
- The Metadata Knowledge Base — what “Build Meta” produces
- Code Generation & Templates — turning metadata into code