The AI team that works while you sleep.
A hierarchy of AI agents on your own hardware — they code, review, and research overnight, then report to your Telegram. Self-hosted, open source, and free forever.
From zero to a running AI team in about ten minutes.
git clone https://github.com/jellyclaw-org/jellyclaw.git
cd jellyclaw
# the core engine ships with the CLI MVP — watch the repoThe one-line installer ships with the CLI MVP. In active development — watch the repo.
One goal in, a whole org chart of agents on it.
Takes your goal, breaks it into department-level work, and keeps the whole team pointed at it.
Plan the work for their area, delegate tasks to workers, and review what comes back.
Execute individual tasks on local Ollama models and report results back up the chain — all the way to your Telegram.
The whole team is one YAML file — the dev-team template looks like this:
# teams/dev-team.yaml — planned format for the CLI MVP
team: dev-team
ceo:
model: qwen2.5:14b # runs locally via Ollama
departments:
engineering:
head: llama3.1:8b
workers: 2
report:
channel: telegramYour office, your hardware, your rules.
Your code and data never leave your machine. Agents run on local Ollama models by default — the only network calls are the ones you configure yourself.
No subscription, no seats, no metering. JellyClaw is open source and runs on hardware you already own. Electricity is the whole bill.
One YAML file defines your whole team. Swap models, restructure departments, or rewire who reports to whom — without touching code.