OpenClaw Operator

Appearance

OpenClaw OperatorBounded, Observable, Auditable AI Operations

Self-hosted operator control plane for real client work, with a built-in console, governed task execution, approvals, run history, incidents, and public proof separation.

Get Started
Run Client Work
Running Agents
OpenClaw Operator lobster mark

GUI-first operator workflow

Run day-to-day work from the built-in `/operator` console instead of wiring raw runtimes together by hand.

Repo-scoped execution

Put target repos inside the workspace tree, point tasks at a bounded scope, and keep edits and evidence reviewable.

Governed execution

Tasks are allowlisted, approval-aware, auditable, and visible through runs, incidents, and health surfaces.

Clear product boundary

Private operator control plane and public proof surfaces stay separate so the runtime can be honest about what is internal and what is public.

Self-hosted deployment paths

Start locally from the root workspace or deploy the full orchestrator stack with Docker Compose.

Knowledge-backed runtime

Local mirrors, knowledge packs, memory recall, and runtime summaries feed downstream operator tasks and agents.

Repo-driven docs

This site is generated from the canonical repo docs so GitHub docs and published docs stay aligned.

Start Here ​

Day-One Usage Model ​

  1. Put the target repo under workspace/projects/<repo-name>
  2. Open /operator/agents to choose the lane
  3. Launch work from /operator/tasks
  4. Read the result in /operator/runs
  5. Use logs/ only for artifact-writing lanes such as knowledge packs or Reddit drafts

Core Paths ​

Docs Site Workflow ​

These published docs are derived from the repository, not maintained as a separate source of truth.

bash
npm run docs:site:dev
npm run docs:site:build
npm run docs:site:preview

Built from the canonical repo docs and generated site source.

OpenClaw Operator