Litmus is an OpenClaw skill that turns your always-on GPU machine into a self-directing ML research lab. Multiple native subagents run experiments overnight, each on their own git branch. A Director steers them every two hours. A Synthesizer distills collective knowledge into a reusable skills library at 04:00. A circadian rhythm shifts agents into creative thinking at 03:00. A research narrative lands in your chat by 08:00.

Built on Karpathy's autoresearch. Everything is a native sessions_spawn subagent — no external CLI processes, no PID files.

Website · Docs · ClawHub

Install by asking your OpenClaw agent:

"Install https://clawhub.ai/kuberwastaken/litmus and set it up for my machine"

Your agent checks your GPU, pitches a full schedule (with timing presets), spawns workers, and registers cron jobs — all in one conversation.

YOU ──▶ OpenClaw agent
             │
             ├── sessions_spawn ──▶ worker-arch-1 ──┐  each on its own git branch
             ├── sessions_spawn ──▶ worker-opt-2  ──┤  in ~/.litmus/repo/
             ├── sessions_spawn ──▶ worker-gen-3  ──┘
             └── sessions_yield
                                        │
                               ~/.litmus/shared/
                                  attempts/       ← JSON record per experiment
                                  notes/          ← structured YAML-frontmatter notes
                                  skills/         ← validated reusable techniques
                                  discoveries.md
                                  anomalies.md

Director (cron · every 2h during research hours)
  └── reads shared/attempts/ → computes improvement rates
  └── Compass Reset on ≥6 experiments without improvement
  └── cross-pollinates discoveries across agents
  └── assigns anomaly investigation

03:00 ── litmus-leisure   ── arxiv scan · contradiction analysis · writes notes/moonshots/
04:00 ── litmus-synthesizer ── distills notes + attempts → skills/ + research agenda
06:00 ── litmus-dawn      ── reads synthesizer output · queues experiments · wakes workers
08:00 ── litmus-digest    ────────────────────────────────────────────────▶ YOUR CHAT

The shared git repo at ~/.litmus/repo/ holds every agent's branch. Browse the full experiment tree any time:

git -C ~/.litmus/repo log --all --oneline --graph

All times are defaults — configurable during onboarding.

Onboarding pitches all defaults and asks what you'd like to change. Common presets:

Full config options saved to ~/.litmus/config.json:

Defaults in configs/default.json. Pass custom times to setup-cron.sh with --leisure-start, --synthesizer-time, --dawn-time, --digest-time, --director-hours, --watchdog-minutes.

# Status — per-agent experiment counts, best val_bpb, stagnation flags, git tree
bash {baseDir}/scripts/status.sh

# Leaderboard — all agents, from shared/attempts/ JSON
bash {baseDir}/scripts/results.sh --top 10
bash {baseDir}/scripts/results.sh --agent arch-1   # single agent

# Full experiment history as a git tree
git -C ~/.litmus/repo log --all --oneline --graph

# Inspect any specific experiment
git -C ~/.litmus/repo show <commit-hash>
cat ~/.litmus/shared/attempts/<hash>.json

# Steer a worker mid-run (no restart needed)
subagents action:"steer" target:"litmus-worker-arch-1"
  message:"Checkout opt-2's best commit and combine their LR with DEPTH=10."

# Stop everything
subagents action:"kill" target:"all"

Or just tell your agent: "How are my Litmus agents doing?" / "Stop all Litmus agents".

• OS: Linux or macOS (Windows not supported)
• GPU: NVIDIA with CUDA
• uv: curl -LsSf https://astral.sh/uv/install.sh | sh
• git: for experiment version control and worktrees
• python3: for JSON attempt records and leaderboard scripts

See INSTALL.md for full details.

MIT-0 — do whatever you want with it.