Clayrune
Mission control for your Claude Code agents.
Run many agents, across every project, and keep working while they do.
Clayrune is a dashboard over the Claude Code sessions you already run. Every
project on one board, every agent named, and the ones waiting on you flagged.
It runs on your machine against your own Claude CLI.
▶ Try the live demo — nothing to install · ⬇ Download — Windows, macOS, Linux
What it does
🎛️ Main features
- 🔒 Runs on your machine against the Claude CLI you are already signed into. No account, no telemetry, MIT.
- 📱 Connect from anywhere in the world using the dedicated mobile app, running over a secure tunnel.
- One dashboard, as many projects as you want. Multi-agent conversations inside each project, with as many running in parallel as you need.
- 🔔 Agents run autonomously until they absolutely need you, then clearly signal they are waiting for an answer.
- Converse with individual agents, each with their own persona, voice and face.
- Hire new agents as you need them through Floor management, where you see everything happening right now.
- 🔀 Multi-vendor and multi-model at the same time: Claude Code first-class, plus Gemini, Codex, Aider, OpenCode and Goose.
- Read output as it streams; interrupt mid-task without killing the session.
- Agents working on the same project are protected by git worktree isolation when you choose it.
Desk connections
Desk's Connections view uses one Add service flow for known services, unknown addresses and user-supplied connection software. Setup precedes enforced Read/Post permissions and passcode Save. Saved-account pickers and GitHub software setup are implemented in the MC-1062/14 follow-up, pending integration and live retesting. Supported saved generation connections offer a human-clicked Check it now from their own tile. See the user guide for limits and information-only suggestions.
🌙 Working without you
- ⏰ Scheduler: once, daily, interval or cron.
- Standing briefs — hand an agent a charter and it sets its own next task.
- 🛡️ In unattended Claude Code and Codex runs, a PreToolUse hook refuses a named list of irreversible shell commands and edits to protected files, unless a Claude Code mod installed on that machine overrules it. It is a list, not a wall: commands it does not name, and MCP tool calls, are not checked.
- Blocked commands and open questions are emailed to you; your reply resumes the same run.
- Starts with the machine, so a 3am reboot doesn't take it offline.
🧠 Memory
- Long-term memory that adapts and keeps learning.
- Agents learn not only from what worked, but from what did not — lessons learned.
- Per-project memory agents read before starting, instead of beginning cold.
- Standing positions, so settled decisions don't get re-proposed.
- Continuous inspection scripts that score how effective your memory files and index actually are.
🧰 Everything else
- Backlogs with priorities, attachments and two-way GitHub Issues sync.
- Skills and MCP servers managed per project or globally.
- Token and cost tracking across sessions.
- Multi-window layouts that survive a refresh and a reboot.
- 🐝 Hivemind mode for fanning out tasks and explorations.
Screens
| |
|---|
| |
The Floor. Every agent currently running, grouped by project, with a face, a name and whichever model it's pinned to. Anything marked NEEDS YOU is stuck waiting on an answer. | Inside a run. You can read what it decided as it goes, and cut in mid-task without killing the session. |
Install
Windows — open PowerShell, paste one line:
iwr https://clayrune.io/install.ps1 -useb | iex
macOS / Linux:
curl -fsSL https://clayrune.io/install.sh | sh
That installs your chosen AI provider CLI, clones the repo and opens the dashboard on
http://localhost:5199. If you'd rather double-click something, use the
Windows download on clayrune.io. The
launcher verifies the exact installer script hash before executing it. Signed
artifacts produced after the Microsoft profile is configured will also carry a
timestamped trusted publisher identity; the current public download is unsigned.
Uninstall
New script-based installs add Uninstall Clayrune to the Windows Start Menu
or ~/Applications/Uninstall Clayrune.command on macOS. The default removes
Clayrune while preserving projects, provider login/state, and a recovery copy
of Clayrune's settings and records. A separately confirmed purge data mode
also removes Clayrune-owned settings, credentials, browser profiles, and local
backups; it still never removes provider CLIs, ~/.claude, or project folders.
The standalone scripts are
installer/uninstall.ps1 and
installer/uninstall-macos.command.
Requires Python 3.9+ and the Claude CLI.
git clone https://github.com/clayrune-io/clayrune.git
cd mission-control
pip install -r requirements.txt
pip install -r requirements-passkeys.txt # optional: passkey support; skip if it fails
python app.py # native desktop window
# or: python server.py → http://localhost:5199
On first run Settings walks you through the port, your projects directory and
which model to use.
Full configuration reference: docs/reference-config-and-features.md.
Common questions
It doesn't replace Claude Code, it runs it. Claude Code is the engine and this
is the cockpit. You'd reach for it once you've got more than one project on the
go, more than one agent, or work you'd rather not sit and babysit.
| Bare Claude Code | Clayrune |
|---|
| One repo, one terminal | Every project, one board |
| You're at the keyboard | Scheduled + unattended agents |
| Session ends, context dies | Cross-session memory |
| Desk-bound | Full control from your phone |
Mostly they don't — the parallelism is across projects, and each project's
agent gets its own directory and its own session.
If two really do need the same repo, switch on Worktree isolation in
Settings (it's off by default). The second agent and any after it get dropped
into their own git worktree on their own branch. Committed work merges back
when the session ends, and if it won't apply cleanly the branch is kept and
you get told, rather than something clever happening behind your back.
What that doesn't fix is two agents thinking about the same file at the same
time. Worktrees stop them overwriting each other's bytes. They don't stop the
second one building on something the first just made untrue. I haven't solved
that.
Multi-provider agents (Claude Code first-class, plus Gemini, Codex, Aider,
OpenCode, Goose) · Hivemind multi-agent orchestration · Scheduler (once / daily /
interval / cron) · autonomous Steward · backlogs with GitHub Issues sync ·
Skills and MCP server management · token and cost tracking · multi-window
layouts that survive a reboot · phone access with push over the clayrune.io
tunnel · plan viewer · shared rules across projects.
Detail: docs/reference-config-and-features.md.
Incidents
Failures we hit building this, written up for people who do not use it. Each one
reported success while doing nothing, which is the only reason they are worth
reading: no error, no red build, nothing in a log.
All of them in docs/incidents.
How it's built
Flask on the back, plain JS on the front with no build step, JSON files on disk
instead of a database, and the claude CLI spawned as a subprocess. One Python
process, one HTML file, no database.
Contributions welcome — fork, branch, run python server.py, check it, open a PR.
Notes in docs/reference-config-and-features.md.
License
MIT, except two source-available-but-proprietary directories:
mc_remote/ and
mc_tunnel/, the closed platform-binding modules for
clayrune.io. The open-core seam is mc_remote_iface/ (MIT) — implement that
contract to plug in your own remote-access provider.
Rationale: docs/remote-access/07-licensing.md.
