Build, iterate, and share reproducible pipelines without the boilerplate.
Visual blocks make data science approachable. Drag, connect, run.
Targeted re-execution - not full notebook reruns. Iterate in seconds, not minutes.
Claude Code, Codex, and other AI agents can build, run, and interpret pipelines programmatically.
Pick from 70+ built-in blocks or write your own Python. Each block is a self-contained step.
Draw edges between blocks to define data flow. Forge handles dependencies automatically.
Change a parameter, hit "Run Stale" - only downstream blocks re-execute. No more full reruns.
Drag a .py file onto the canvas. Dependencies install automatically. Organize blocks into custom categories with their own colors and icons. Export, share, and version-control your plugins.
Hover, zoom, pan, and inspect results inline. Plotly charts run as blocks like everything else - change a parameter upstream and the chart updates on its own.
Wrangle messy CSVs into structured datasets with reusable blocks.
Train, tune, and compare models with visible data flow at every step.
Prototype fast. See intermediate results without re-running everything.
Save and share pipelines as JSON. Same inputs, same outputs, every time.
Extend Forge with your own code. Full Python - no sandbox, no limits.
Scatter plots, heatmaps, bar charts, histograms - 11 built-in plot types, plus 3D.
Forge is free and open source under the Apache 2.0 license. Built by a data scientist, for data scientists.
Thanks to Nick Robinson for early contributions, including Plotly support.
Premium features (team collaboration, cloud execution) coming soon.
