DomBot LabOS — deploy your own autonomous AI agent infrastructure.
Hub, Kanban, multi-agent routing, shared memory. On your machine.
Autonomous AI agents are powerful — but they're all black boxes.
LabOS is different: every decision is visible.
Confidence score, agent selected, cost, reason — all logged before execution.
Self-hosted. Vendor-free. Open source.
Simulated — real interfaces running on your self-hosted instance
| Time | Agent | Action | Confidence | Message |
|---|
Unified dashboard with real-time SSE stream, system health metrics, and quick actions for all your lab services.
Task board with Kahneman-inspired confidence scoring (0–1). Dependency enforcement, audit trail, agent feedback loop.
Orchestrate OpenClaw, Claude Code, Gemini, and Mistral. Auto-select by cost, quality, or speed — per task.
Reusable workflow packages. TDD, brainstorming, code review, debugging — pre-configured, community-contributed.
Shared Obsidian vault with L1/L2/L3 architecture. Persistent agent memory across sessions and reboots.
Runs on a VPS or Raspberry Pi 4. No third-party cloud required. Your data stays on your machine.
Every agent action is logged with a confidence score before execution. Hub displays real-time updates via SSE. Immutable audit trail.
Workflow engine coordinates multiple agents across sequential or parallel steps. Router selects cheapest agent meeting quality threshold. Skills are reusable across projects.
git clone https://github.com/lgiron/dombot-labos
cd dombot-laboscp .env.example .env
# Add CLAUDE_API_KEY (required)
# MISTRAL_API_KEY, GEMINI_API_KEY (optional)docker-compose up -d
# Dashboard at http://localhost:8088Two-layer system design, agent adapters, confidence scoring.
Writing, using, and publishing reusable agent workflows.
Kanban API, Logs API, agent adapter interface.
VPS, Raspberry Pi, Docker, Nginx config, Authelia auth.
Agent identity, roles, network policy, audit trail.
Phase 1–6 plan, current status, community contributions.