| name | marimo-builder |
| description | Create, edit, and deploy marimo reactive notebooks. Mount notebooks as FastAPI endpoints for interactive dashboards and control surfaces.
|
| version | 0.1.0 |
| author | Jero (LATTICE / MARPA Design Studios) |
| triggers | ["create a marimo notebook","marimo dashboard","mount notebook as API","reactive notebook","control surface"] |
| tools | ["Bash","Read","Write","Edit","Glob","Grep"] |
Marimo Builder
USE WHEN the user wants to create or edit marimo notebooks, build reactive dashboards, mount notebooks as FastAPI endpoints, or create control surfaces for GRAFTKIT operations.
What It Does
Build and deploy marimo reactive notebooks as part of the GRAFTKIT stack:
- Create — Generate new marimo notebooks with PEP 723 inline metadata and proper cell structure
- Edit — Modify existing notebooks, add cells, wire reactive dependencies
- Mount — Serve notebooks as FastAPI endpoints via
marimo.create_asgi_app()
- Connect — Wire notebooks to PostgreSQL, Pixeltable, and events.jsonl data sources
- Deploy — Package notebooks for production serving
Usage
create a marimo notebook for monitoring agent activity
mount the dashboard notebook on /dashboard
add a SQL cell to the existing notebook
Notebook Structure
import marimo
app = marimo.App(width="full", app_title="Title")
@app.cell
def cell_name():
import marimo as mo
return (mo,)
FastAPI Mounting
from fastapi import FastAPI
import marimo
app = FastAPI()
marimo_app = marimo.create_asgi_app()
marimo_app.include_app("/dashboard", "notebooks/dashboard.py")
app.mount("/", marimo_app.build())
Data Sources
| Source | Connection | Use For |
|---|
| PostgreSQL | psycopg2.connect(dbname="graftkit") | Table stats, rootstock queue |
| Pixeltable | import pixeltable as pxt | Multimodal data, embeddings |
| events.jsonl | Path("logs/events.jsonl") | Pipeline events, tool usage |
| library.yaml | yaml.safe_load() | Engraft catalog status |