Run multiple autonomous ML research agents on GPU machines with git branches.
npx skills add https://github.com/Kuberwastaken/litmus --skill litmus
Please help me install this Agent Skill. Skill: litmus Source: https://github.com/Kuberwastaken/litmus/tree/main Command: npx skills add https://github.com/Kuberwastaken/litmus --skill litmus
Litmus turns your always-on GPU machine into a self-directing ML research lab where multiple native subagents run experiments overnight, each on their own git branch, coordinated by a Director and distilled by a Synthesizer that builds a reusable skills library.
Install Litmus on your OpenClaw agent and let it configure defaults and spawn workers automatically.
High-intent search queries and answers about installing and using this skill.
Running autonomous ML research overnight requires a GPU-equipped machine with OpenClaw, uv, git, and python3 installed. The Litmus Skill orchestrates parallel agents that independently run experiments on dedicated git branches throughout the night.
To set up parallel ML research agents, you need an always-on GPU machine with OpenClaw, uv, git, and python3 installed. Litmus configures default settings and automatically spawns worker agents upon installation.
Autonomous ML research coordination uses a Director to steer progress and trigger compass resets on stagnation. Shared lab state enables cross-agent knowledge transfer, allowing parallel experiments to collaborate and share insights.
Overnight ML experiment results are distilled by a Synthesizer into a reusable Skills Library and a forward-looking research agenda. A morning digest delivers a narrative summary of discoveries, patterns, and next steps for human review.
Tracking individual ML experiment history is possible because each autonomous agent runs on its own git branch with full commit history. This isolated branching allows you to review and cherry-pick breakthroughs from specific experiments.
saiteja007-mvCoordinate autonomous AI research workflows on GPU-enabled devices.
goffityExtracts patterns and insights from linked notes to create reusable knowledge base entries.
julianobarbosaManage Git history across branches with rebasing, cherry-picking, and reflog recovery.
x-TheFoxLog ML experiments, sweeps, and artifacts with WandB.
