shabadoo

Your agent stopped to ask a question
twenty minutes ago.

Claude Code sessions run for a long time and then stop dead on a prompt โ€” a permission dialog, a plan approval, a yes/no. You walk away, come back, and find it spent nineteen of those minutes waiting for you to press Enter.

shabadoo is a self-hosted control plane for every Claude Code session on every machine you use. One dashboard. A queue of the ones waiting on you. A notification when one has been stuck. And the ability to answer from wherever you are.

The bottleneck is not the model

It is the round trip back to your attention. Multiply one blocked session by the number of projects you have going, across a workstation and a laptop and a box under the desk, and the cost is not compute โ€” it is you, not being there.

It knows when a session blocks

Each host's agent classifies every pane. A session waiting on a prompt is flagged the moment it happens โ€” no polling from your browser.

It tells you

After 90 seconds of a standing prompt, a notification reaches your phone. Long enough that the ones you answer immediately never buzz.

You answer from anywhere

Tap through to the pane, read what it is actually asking, send the keystroke. From a browser or a phone, over your own private network.

How it works

One static binary, three roles. A coordinator you run somewhere always-on, an agent on each machine that has tmux, and the launcher that starts the sessions in the first place.

# on the always-on box
shabadoo hub --device-tokens --addr 0.0.0.0:8787

# on every machine you actually work on
shabadoo setup --service --coord https://coordinator.example

# then, from anywhere
shabadoo sessions            # every session on every host
shabadoo tail homelab        # what is that pane asking?
shabadoo keys --window 3 Enter

Agents dial out

Your laptop needs no inbound port, no tunnel, no static address. It works behind NAT, on hotel wifi, asleep half the day. Only the coordinator has to be reachable.

One binary, no dependencies

Copy a single file to a fresh machine and it installs the services, the launcher and the config โ€” with no network and no source tree. No Node, no Python, no Docker required.

Sessions talk to each other

Each session is the expert in its own project. They message each other by domain โ€” send to "homelab" โ€” see who is online, and say what they are working on.

Everything is audited

Every keystroke sent into every pane is recorded and attributed to an enrolled device. Read-only enrolment exists for anything that only needs to watch.

Many sessions, working as one

The premise is not a web terminal. It is that you already run several Claude sessions, each holding context nobody else has, and they should be able to act as one system โ€” monitored and driven from anywhere, without any of them leaving your network.

CapabilityWhat it does
Route by domain Hand a task to whichever session owns a project. Ambiguous names are refused rather than guessed; unknown ones bounce with the list of what exists.
Durable inbox Mail waits for a session that is offline and is delivered when it returns. Draining is final, so nothing is silently redelivered.
Declared status tmux can tell you a window is idle. Only the session can tell you it is idle because it is waiting on a peer.
Blocked-session queue Everything waiting on a human, longest first, at the top of the dashboard โ€” rendered only when it is not empty.

Read this before you run it

Not fine print. If any of this is not acceptable to you, this is the wrong tool and it is better to find that out here.

Reaching the dashboard means driving every pane.

Whoever holds a credential can read every project path, read the full buffer of any pane, read any Claude session transcript, and send keystrokes into any pane. Those panes typically run claude --dangerously-skip-permissions, so a keystroke can do anything Claude can do on that machine.

Read-only enrolment limits a client to watching, and is the right default for anything that does not need to type. It mitigates this; it does not eliminate it.

The coordinator concentrates that reach, deliberately.

Authorising an agent is the same decision as handing that machine's Claude panes to anyone who can reach your coordinator. Run it on a private network โ€” a VPN or a tailnet โ€” and never expose it to the internet.

Everything is self-hosted. There is no cloud service, no account, and no telemetry: your prompts, panes and transcripts never leave your machines.

What it is not

LimitationDetail
Requires tmux Sessions must run in tmux. If you use Claude Code in an IDE or a bare terminal, this is not for you.
macOS and Linux No native Windows agent. WSL works.
One user Multi-tenancy is implemented but barely exercised. Treat it as single-operator.
Notifications via a relay Blocked-session alerts go through an Apprise relay to Telegram, Pushover and similar. Native iOS push is in progress.