LabLoop
github.com/cloudcell/labloop- Category
- Developer Tools
- Rank
- No. 1747Tools index
Previous survey · No. 1753 ·
- Pricing
- Open Source
- Type
- TOOL
- Use case
- Research & Education
- Interfaces
- MCP
- GitHub
- 8 stars
- Latest release
- v0.0.2
- Date
About
LabLoop is a research automation environment that pairs ml-labloop, a KVM-based isolation harness running per-user VMs split into driver, trusted, and hostile zones, with ml-scientist, a set of five MCP servers (agora, arete, zetesis, episteme, anamnesis) that record hypotheses, experiments, evidence, and conclusions as durable state. This lets AI agents run untrusted experimental code inside sealed VMs while keeping a reproducible scientific record separate from the code execution.
At a glance
LabLoop is the former name of a research system that records hypotheses, experiments and evidence for AI agents. Start with GnosisLab for the research tools, or choose GnosisLab-VM when you also need a virtual machine for experiments.
Good fit
Research teams that want another person or agent to trace a conclusion back to the experiment and evidence that produced it.
Consider something else when
A notebook and versioned results may be enough for a small experiment with no need for an agent to query prior work.
Start here
Use the current GnosisLab documentation to start the services and run gnosislab status. Record one hypothesis and a small synthetic experiment, then check whether you can trace an observation back to its trial.
Cost and requirements
Both successor projects use Apache 2.0. The standalone runtime targets Python 3.10 or later. Building and managing the VM with the documented KVM tools requires a Linux host with virtualization support.
Limits to know
The LabLoop repository retains historical commands. Use the successor documentation for setup and check its execution isolation requirements before running untrusted experiments.
Sources checked 2026-11-05. Source-based assessment.
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.