Claw Space is a local mission-control UI for an OpenClaw workspace. It runs on your machine, talks to the openclaw CLI, and gives you a browser view for chat, tasks, sessions, memory files, agent settings, and swarm-style work planning.
It is built for loopback use on a developer machine. It is not a hosted dashboard, and it should not be exposed to the public internet without adding your own authentication layer.
- A dashboard for agent health, recent sessions, task counts, and workspace activity.
- Agent chat with streaming output, session continuity, and file attachments.
- A Kanban board for local task queueing and one-click task runs.
- Session log browsing, previews, exports, and memory file editing.
- Agent settings views for identity, model, thinking level, fast mode, skills, and tools.
- A swarm board for splitting a goal into worker lanes and tracking each assignment.
- Mock mode for UI development without calling real agents.
The dashboard keeps the important local state in one place: configured agents, recent session logs, task counts, and whether the server is running in loopback-safe mode.
The swarm board is for breaking a larger goal into worker lanes, reviewing assignments, and keeping runtime output visible while work moves through the board.
Requirements:
- Node.js 20 or newer.
- OpenClaw installed locally.
- An OpenClaw config, usually
~/.openclaw/openclaw.json.
git clone https://github.com/Martinnn674/claw-space.git
cd claw-space
npm install
npm startOpen:
http://127.0.0.1:8787
If openclaw is not on your PATH, point the UI at it:
OPENCLAW_BIN=/path/to/openclaw npm startFor UI work without a real OpenClaw install:
OPENCLAW_WEB_UI_MOCK=1 npm start- Setup guide
- Showcase
- Feature tour
- Development notes
- Maintainer guide
- Privacy checklist
- Roadmap
- Changelog
The app reads local OpenClaw memory, session logs, and config-adjacent data. Keep it on 127.0.0.1 unless you know exactly what you are doing.
Before releases, run:
npm run checkThat includes syntax checks, the tracked-file privacy audit, the mock smoke test, and the browser smoke test.
Early, usable, and intentionally small. The current focus is keeping local agent work easier to inspect without turning OpenClaw into a cloud service. Maintenance work is tracked through GitHub issues, the roadmap, and release notes.
The UI incorporates MIT-licensed visual patterns from Hermes Workspace. See THIRD_PARTY_NOTICES.md.