Open Science is an open-source AI workbench built for researchers who want the power of an AI agent — without vendor lock-in, data leaving their machine, or a subscription fee. It pairs a model-agnostic agent runtime with a persistent Python kernel, durable artifact storage, and rich previews for scientific file formats including CSV, FASTA, PDB, and Markdown.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/aipoch/open-science/llms.txt
Use this file to discover all available pages before exploring further.
Quickstart
Install Open Science and run your first agent-assisted analysis in minutes.
Installation
Download pre-built binaries for macOS, Windows, and Linux, or build from source.
Core Concepts
Understand the architecture: agent harness, notebook kernel, artifacts, and MCP servers.
IPC API Reference
Explore the full
window.api surface exposed to the renderer via Electron’s preload bridge.What Open Science Does
Open Science provides a plan → execute → produce → preview loop for scientific work:- Plan — describe a research task in the conversation composer
- Execute — the AI agent writes and runs Python code in a persistent kernel
- Produce — outputs are stored as versioned artifacts in your project
- Preview — results render inline: tables, plots, sequences, structures, and more
Workspace Guide
Learn the three-panel workspace: sidebar, conversation, and preview.
Notebook Kernel
Understand how the persistent Python kernel executes agent-generated code.
Artifact Storage
Learn how outputs are stored, previewed, and organized per session and run.
File Previews
Preview CSV, FASTA, HTML, JSON, Markdown, PDB, images, and source files inline.
Why Open Science
Model-Agnostic
Designed from the ground up to support any AI model — not locked to a single provider.
Local-First
All data stays on your machine. No cloud sync, no telemetry, no subscription.
Reproducible by Default
Every agent run records its code, outputs, and artifacts with run-level provenance.
Open Source
Apache-2.0 licensed. Inspect, modify, and contribute on GitHub.
Open Science is in Early Alpha. The core plan → execute → produce → preview loop works end-to-end. Some planned features — model-agnostic gateway, artifact versioning, R kernel, skills commons — are on the roadmap. See the Roadmap for details.