Science Without Walls
Local-first, model-agnostic AI research workbench. Run the entire research loop — from exploration to publication — in one auditable desktop session.
From Idea to Paper, Autonomously
One meta-skill drives four stages — each produces real, inspectable artifacts, not just chat replies.
Explore
Turn a broad research direction into concrete topics, hypotheses, and a pre-survey.
Survey
Produce a 6–20 page PDF literature review with 60+ real, verified citations.
Experiment
Design docs, runnable code, results with provenance, figures, and analysis reports.
Write
8–14 page research paper, 200+ citations, 4–8 figures — publication-grade output.
Built for Real Research
Every feature designed to accelerate scientific work while keeping you in full control.
Autonomous Research Agents
The bundled ai4s-agent chains specialist skills end to end — explore, survey, experiment, write — producing inspectable artifacts at every stage.
Full Provenance Tracking
Every figure, table, and report links back to the exact code, inputs, environment, model output, and conversation that produced it.
Local-First Architecture
Sessions, data, provenance, notebooks, and run records live in local folders on your machine. Nothing leaves by default.
Model-Agnostic Runtime
Bring your own model. The runtime talks through a clean SDK layer — providers, skills, and MCP servers stay pluggable.
Reproducible Runs
Local, SSH/Slurm, Modal, and notebook-batch runs are captured as reproducible run records with full environment snapshots.
Access from Anywhere
A token-authenticated gateway serves the real UI to your phone or LAN browser. Start a run at your desk, check results on the go.
Browser Control
Drive your own Chrome — profile and logins intact — to read the live web the way you do, or an isolated private browser on demand.
Highly Extensible
Agent skills, MCP servers, one-click science connectors (arXiv, PubMed, Crossref, Materials Project…), slash commands, and shell mode.
Powerful Research Toolkit
A comprehensive platform that covers the entire scientific workflow.
Multi-Session Workspace
Run multiple research sessions simultaneously with isolated workspaces, shared file system, and full history.
Integrated Notebooks
Real .ipynb files with Python and R kernel execution, managed Jupyter environment, and seamless agent integration.
Rich Viewers
PDF, images, video, HTML, Markdown, CSV/TSV tables with charts, molecules, 3D meshes, genome tracks, and more.
Literature Connectors
One-click access to arXiv, PubMed, Crossref, Semantic Scholar, bioRxiv/medRxiv, and more research databases.
Integrity Auditing
Built-in auditor checks images, numbers, logic, and citations with 4-level evidence grading before publication.
Safety by Default
Command execution, file deletion, dependency install, and remote connections all require explicit human approval.
The Most Complete Open Research Desktop
Where other AI-for-science tools trade away platforms, models, or openness, ZeroWall keeps all of it.
True Cross-Platform
Prebuilt installers for macOS, Windows, and Linux — the only tool in its class with no missing OS and no mandatory source build.
Reach It From Your Phone
A token-authenticated gateway serves the real desktop UI to your phone or LAN browser. No other project here offers mobile access.
Every Field, Not Just Biology
General-purpose across biomedical, materials, climate, economics, and astronomy — not locked to a single domain.
Open, Model-Agnostic, Free
MIT-licensed, bring any model you like, no account or subscription. You own the whole workflow end to end.
How ZeroWall Compares
An honest, side-by-side look at ZeroWall Science and other AI-for-science tools.
| ★ This project ZeroWall Science | Claude Science | Biomni | wisp-science | |
|---|---|---|---|---|
| Platforms | ✓ macOS · Windows · Linux | ✕ macOS · Linux only (beta) | ◐ macOS · Linux via conda | ◐ Win · macOS; Linux source-only |
| Login & setup | ✓ One-click account login · one-click model keys · zero config | ✕ Paid Claude subscription login | ✕ pip install + 11 GB download + manual keys | ◐ Manual API-key configuration |
| Research domain | ✓ General science, many fields | ◐ Biomedical focus | ✕ Biomedical only | ◐ Bioinformatics focus |
| Model choice | ✓ Model-agnostic, any provider | ✕ Claude models only | ✓ Model-agnostic | ✓ Model-agnostic (no Gemini) |
| Bundled agent skills | ✓ 70+ skills: explore · survey · experiment · write + life-science | ✓ Built-in Agent Skills (biomedical) | ✓ ~150 tools (biomedical only) | ◐ Small built-in set (MVP) |
| Science connectors (MCP) | ✓ 23 one-click: arXiv · PubMed · ChEMBL · trials… | ◐ Built-in literature / data retrieval | ✓ ~59 databases (biomedical) | ◐ Few, manual wiring |
| Integrated notebooks | ✓ Real .ipynb · Python + R kernels | ◐ Files + code execution | ◐ Code execution · Gradio (no notebook UI) | ✓ Persistent Python + R REPL |
| Provenance tracking | ✓ Full chain: code · data · env · model | ◐ File versioning + review | ✕ No systematic provenance | ◐ Basic run records |
| License | ✓ Open source · MIT | ✕ Proprietary, closed | ◐ Apache-2.0 (frozen 2025-04) | ✓ Open source · Apache-2.0 |
| Local-first & private | ✓ Yes — nothing leaves by default | ◐ Local files, cloud inference | ◐ Local run, 11 GB data download | ✓ Yes — SQLite + OS keychain |
| Safety approvals | ✓ Exec / delete / install / remote | ✓ Folder / host / job approvals | ✕ Runs code with full privileges | ◐ Probe-gated remote compute |
| Price | ✓ Free | ✕ Paid plan required | ✓ Free (library) | ✓ Free |
Compiled from public docs and source, July 2026. Competitors are independent projects; this is a neutral feature comparison, not an endorsement.
Clean Layered Design
Decoupled layers with a strict SDK boundary — the UI never calls the agent runtime directly.
Get ZeroWall Science
Free, open-source, available for all major platforms.
macOS
Apple Silicon & Intel, macOS 13+
Coming SoonLinux
x86_64 .deb / .rpm
Coming Soon